Introduction

The prevalence of obesity is constantly rising, creating a major global health problem [1]. There are numerous studies and meta-analyses which demonstrate that obesity is related with several comorbidities such as type 2 diabetes mellitus (T2DM), cardiovascular disease (CVD), heart failure (HF), hypertension, and various types of cancers [2]. We refer to the obesity paradox when particular chronic diseases, that will be reviewed hereunder, exhibit an interesting “paradoxical” protective association between body mass index (BMI) and clinical outcomes. Nevertheless, in the absence of coexisting conditions with may display these paradoxical associations, body weight has unequivocally been shown to display a J-shaped association with multiple causes of morbidity and mortality as well as with the leading causes of death globally, especially in the setting of obesity classes II and III [3, 4].

Overweight and obesity are generally defined using the body mass index, both in clinical practice and in epidemiological studies [5]. The need of an index describing the magnitude of adiposity was originally generated in 1832 by a Belgian mathematician, statistician, and astronomer named Quetelet who used the ratio of the weight in kilograms divided by the square of the height in meters. This index was validated by Ancel Keys in 1972 naming it body mass index (BMI) [6]. The World Health Organization defines overweight and obesity as abnormal or excess body adiposity that may impair health, underlining that BMI may not reflect the degree of fat percentage among different individuals [7]. Therefore, it is essential to include additional parameters in order to correctly define overweight and obesity, as a trait which reflects the sum of adiposity, such as waist and hip circumferences and, when feasible, assessments of fat/muscle percentage via body composition analysis [8]. Anthropometric and body composition analysis could showcase individuals with similar BMIs, but different body composition data in terms of lean and fat mass, thus differentiating health outcomes [9]. Furthermore, several medical conditions have been associated with low muscle mass and function or with the coexistence of reduced lean and increased fat mass, notably in the elderly population, a phenomenon known as sarcopenia or sarcopenic obesity, respectively [10, 11]. These observations could partly set the ground for understanding how the term “obesity paradox” was generated. Consequently, in order to clarify the controversial issues around obesity related outcomes, BMI determination does not suffice, and additional indices resulting from the quantification of different body compartments such as fat and lean mass are of utmost importance [12].

The obesity paradox has been described in a plethora of chronic diseases. That said, it is important to note that despite the literature pointing towards an obesity paradox, there is very strong evidence supporting obesity’s role in the pathogenesis of multiple diseases including CVD and is associated with reduced overall survival. The paradoxical “protective” effect of overweight or mild obesity has been reported mainly in CVD, including coronary heart disease (CHD), heart failure (HF), hypertension, and atrial fibrillation (AF) [13]. However, this paradoxical effect has appeared in several other conditions such as cancers, chronic kidney disease, chronic obstructive pulmonary disease (COPD), stroke, pulmonary hypertension, osteoporosis, critical illness, and sepsis (Fig. 1) [4, 14,15,16,17,18,19,20]. It cannot be emphasized enough that the incomplete information received from a single BMI measurement may interfere with outcomes of studies arguing in favor of the obesity paradox. A large meta-analysis of 10,625,411 participants that included 239 prospective studies from 4 continents showed that the relationship between BMI and all-cause mortality is generally J-shaped. This could point to the fact that if underweight were to be excluded from this analysis, the association of higher body weight with mortality would be even stronger. The nadir BMI for low mortality varied with age with participants aged 35–49 years old exhibiting a nadir mean BMI of 22 kg/m2. Participants aged 50–69 years old had a mean BMI equal to 23 kg/m2, and those aged 70–89 years old were found with a mean BMI of 24 kg/m2 [21]. Another meta-analysis of 230 cohort studies demonstrated that the BMI-associated mortality curve is U-shaped with an elevated mortality at the extremes of obesity and underweight among all participants. However, a J-shaped curve was observed among participants that were never smokers [22]. A large cohort study demonstrated that in middle aged-adults were smokers, obesity is associated with shorter longevity and increased cardiovascular mortality compared with normal weight patients. Of note, 48% of normal weight patients were smokers versus 32,7% in patients with obesity [23].

Fig. 1
figure 1

Conditions associated with the obesity paradox

Full size image

BMI might be a very easy, costless method to classify adiposity; however, there are various different obesity phenotypes, suggesting that people with similar BMIs may present with different body compositions. A first such example is the nonsarcopenic phenotype of obesity with both increased fat and lean mass. Then, there is the athlete’s phenotype with obesity, i.e., decreased fat and increased lean mass, and lastly, the sarcopenic phenotype of obesity with elevated fat and reduced lean mass [24, 25]. A substantial number of studies have shown that there were different health outcomes among individuals with the same BMI, but different body composition data, since BMI does not account for the degree or site of adiposity [12]. A prospective cohort study involving 10,265 men demonstrated that there is an inverse and independent association between muscular strength and all-cause mortality. After adjustment for age, the authors showed that the group with high muscular strength and cardiorespiratory fitness had significantly 60% lower mortality rates than the opposite group [26].

The concept of metabolically healthy obesity has also recently emerged, in which adults with obesity do not present with metabolic comorbidities and ultimately may have lower mortality than leaner individuals with cardiovascular risk factors [27] or may have a higher level of fitness [28]. One should consider the heterogeneous nature of obesity and thus should possibly regard metabolically healthy obesity as a transient state [27] which suggests that people living with obesity are at increased risk for adverse long-term outcomes even in the absence of metabolic abnormalities, i.e., no healthy pattern of obesity actually exists [29].

Methodological and study limitations may also interfere with the association of obesity with morbidity and mortality, originating from the lack of understanding that obesity is a chronic disease [30].

Obesity year exposure may play an important role. Relating obesity duration to cardiometabolic disease risk factors in mid-adulthood has shown that a larger obesity duration exposure is associated with worsening of cardiometabolic disease risk factors, although this positive association was attenuated when adjusted for obesity severity [31]. Another possible explanation includes the bias due to the so called reverse causation where antecedent weight loss due to chronic illness like cancer may elevate mortality risk [32]. It has been proposed that excess adipose tissue may act as a metabolic reservoir, in a way protecting exposure of additional tissues to lipotoxicity and ectopic fat deposition along with acting as a buffer for dietary fat influx [33]. Several of the aforementioned confounding factors, i.e., the subjects’ underlying health status, antecedent body weight loss due to chronic illness or ageing, and smoking, may on several occasions be challenging to surmount, and thus, caution should be exercised in interpreting study results [32]. The aim of the present review is to examine the controversial issues around the obesity paradox in certain conditions such as cardiovascular disease, several types of cancer and chronic obstructive pulmonary disease, and the factors that may confound the relation between obesity and mortality.

Obesity Paradox in Cardiovascular Disease

Coronary Heart Disease

The association of obesity with increased mortality risk has long been established [34], and multiple studies have pointed out that obesity has a detrimental effect on CVD. Pioneering research in the field was performed by Ellis et al., who tested the hypothesis that individuals with both BMI < 20 kg/m2 and BMI > 35 kg/m2 would possibly demonstrate increased mortality risk after percutaneous coronary intervention (PCI). They enrolled 3571 patients from a single center with a median follow-up of 1 year, and their initial hypothesis was found to be correct [35]. Diletti et al. enrolled 5127 patients with CAD and followed them up for 2 years. The analysis revealed that after drug-eluting stent implantation BMI had no association with mortality [36]. Another small cohort study observed the death rates after PCI and concluded that there were no significant differences after 30 days among all BMI groups [37]. Akin et al. also reported no obesity paradox after 1 year of follow-up in patients who underwent PCI with drug-eluting stents [38].

On the other hand, and in favor of the obesity paradox, there are multiple studies in the literature. In a cohort of 3634 patients, a subgroup of 1829 was randomized to either coronary artery bypass graft surgery (CABG) or percutaneous transluminal coronary angioplasty (PTCA) examining the short- and long-term outcomes after initial revascularization. Interestingly, with each unit increase of BMI, there was a 5.5% lower adjusted risk of a major in-hospital event after PTCA, whereas no significant differences were observed after CABG. In the long term, the mortality rates were lower among PTCA-treated patients with overweight or class I obesity compared with normal weight patients. In the CABG group, the 5-year mortality risk depicted a positive linear correlation with class II/III obesity [39]. The observation that in-hospital complications after PCI were higher in underweight and normal weight patients was also confirmed in a study of 9633 patients [40]. Additionally, a different cohort demonstrated that lean patients had the highest in-hospital mortality risk even though the multivariate analysis showed that BMI had no impact on it [41]. In agreement with the aforementioned came a large Swedish study in which the association between BMI and all-cause mortality was U-shaped in patients who underwent coronary angiography due to acute coronary syndromes, with the lowest mortality noted to be when BMI was between 30 and 40 kg/m2. Notably, the unadjusted survival in people with CAD was lower when BMI was under 18 kg/m2 as demonstrated by the Kaplan-Meir curve [42]. A Korean study including 3824 patients who underwent PCI after a ST-segment elevation myocardial infarction (STEMI) showed that serum N-terminal pro-brain natriuretic peptide (NT-pro-BNP) levels were significantly higher in underweight and normal weight patients than their counterparts with overweight and obesity. These laboratory findings suggested hemodynamic instability, and in combination with infrequent use of β-blockers and angiotensin-converting enzyme inhibitors, lean and underweight patients had significantly higher mortality risk [43].

However, there are many essential variables that need to be examined in order to determine whether obesity is actually responsible for better clinical outcomes in patients with CAD. Data from a multicenter, multinational population with CAD demonstrated that from 16 potential variables for cardiovascular mortality, the independent variables included age, BMI, current smoking habits, diabetes mellitus, total cholesterol levels, and history of myocardial infarction, congestive heart failure, revascularization, or stroke [44]. Numerous studies have shown that patients with obesity and CAD are more likely to be younger than lean patients with CAD [45]. Oreopoulos et al. examined 31,021 patients for a median follow-up of 46 months and showed that patients with obesity who had undergone PCI or CABG were younger and most likely never smokers. Patients with class I obesity in the CABG group and patients with class II obesity in the PCI group had the lowest mortality risk [45]. There is evidence suggesting that advanced age is related to worse health outcomes independent of course of treatment [46]. On the same note, substantial research in 409 US hospitals revealed that patients with obesity and CAD were younger and had more intensive guideline focused therapy amongst more comorbidities. Thereby, greater BMI was associated with increased use of aspirin, lipid lowering drugs, b-blockers, and antihypertensive medications both at hospitalization and in the long term [47]. Adiponectin might contribute to this paradoxical effect, thus proposing a different paradox [48]. Naturally, adiponectin is secreted by adipocytes, and its levels are decreased in people with obesity [49]. Whereas adiponectin is known, among others, for its cardioprotective effects, recent data have demonstrated the “adiponectin paradox,” wherein people with established CVD and elevated adiponectin had a greater mortality risk [50]. There is evidence indicating that elevated adiponectin levels in CAD are mediated by atrial natriuretic peptide (ANP) and brain natriuretic peptide (BNP) [48]. Thus, this association may not in fact be causal.

Obesity is well known for its low-grade subclinical inflammatory state. Systemic inflammation is characterized by high-sensitivity C-reactive protein (hsCRP) levels, and a large number of studies have implicated its role in CVD. Anti-inflammatory therapy has been shown to reduce hsCRP levels and the risk for a major cardiovascular event or death [51]. On the other hand, there have been studies suggesting that mortality was significantly lower after cardiac rehabilitation in a study population that had high CRP and low body fat. The authors discussed the protective role that elevated CRP may have, or the increased lean mass, or the combination of both [52].

Heart Failure

By definition, the obesity paradox in heart failure (HF) indicates towards low mortality rates in patients with overweight or obesity. People with obesity are prone to be more often diagnosed with HF, but it appears that obesity exerts a protective effect after diagnosis [53]. Indeed, Curtis et al. studied 7767 patients with stable HF for 37 months and highlighted that increased BMI was associated with lower mortality risk [54]. Patients with a BMI over 45 kg/m2 presented with increased mortality rates, formulating a U-shaped relationship between mortality and BMI [55]. The association between BMI and mortality was also present in patients with chronic HF with both reduced ejection fraction (EF) and preserved EF (HfrEF, HfpEF) [56]. Additionally, a study investigating patients with both ischemic HF and non-ischemic HF showed that the paradox association between BMI and mortality was present only in patients with non-ischemic HF. The medical history of the non-ischemic patients included higher prevalence of AF, worse NYHA class, and a majority of women, whereas the ischemic patients had higher prevalence of diabetes, hypercholesterolemia, and peripheral vascular disease [57]. Interestingly, in order to further elucidate the obesity paradox phenomenon, the relationship of body surface area was investigated in patients with HF. The investigators highlighted that for each 0.1 m2 increase in body surface, there was an inverse association with mortality, but the aforementioned association did not apply to hospitalizations [58]. Specifically, a study enrolled 3,811 patients with HF and EF < 40% and found that the unadjusted analysis was in favor of the obesity paradox. On the contrary, after adjustments for several confounders, the results revealed that this paradox was present only in females, while the male population had a significantly increased mortality rate when overweight or obesity were present [59]. Notwithstanding, the obesity paradox was also confirmed by a Spanish study, and the authors stressed that the protective effect was lost when waist circumference was over 120 cm [60]. HF is characterized by a chronic catabolic state and in the long-term cardiac cachexia might occur, accompanied with high mortality risk [61]. To wit, people with obesity can cope better with the catabolic state due to higher metabolic reserves. In an attempt to assess the nutritional status in patients with HF, a couple of interesting studies found that undernourished patients had the highest mortality rates [62, 63]. Moreover, patients with HF and obesity seem to have increased prevalence of co-existing medical conditions like T2DM or hypertension. Those findings suggest that the more frequent use of β-blockers and antihypertensive drugs such as ACE inhibitors may play a favorable role in mortality rate [64]. ACE inhibitors reduce mortality in patients with HF, preventing deaths especially to those with high levels of catecholamines [65]. Another confounder in the obesity paradox is the greater prevalence of non-smokers or ex-smokers, as well as a lower incidence of previous STEMI or non STEMI in patients with obesity [64].

We mentioned previously that NT-proBNP and BNP levels were significantly higher in people with obesity and CAD. BNP is a neurohormone secreted by the cardiac ventricles, expressing pressure overload and volume expansion [66]. Mehra et al. investigated people with congestive heart failure, measuring BNP levels, and confirmed that people with obesity had significantly lower levels and thereby better prognosis. In order to exclude the age confounder, they divided the subjects to elderly and nonelderly, with no difference in the results [67].

Atrial Fibrillation

Several studies agree on the increased prevalence of atrial fibrillation among patients with obesity [68]. What is intriguing though is that in the last decade, an ample amount of research manifested a similar paradox among people with obesity and atrial fibrillation as described in CAD and HF. In a large study of 17,913 patients with AF, patients were randomized to receive warfarin or apixaban and were followed for 1.8 years. The multivariate analysis demonstrated a lower risk for all-cause mortality and significantly reduced adverse events like stroke, bleeding, MI, or systemic embolism in people with overweight and obesity compared with normal weight patients. Additionally, the patients that were randomized to the warfarin arm had increased incidence in all complications investigated compared to the apixaban arm. Patients in the apixaban subgroup with a normal BMI had lower prevalence of major bleeding and stroke compared with higher BMIs [69]. The paradox association between BMI and AF was also apparent in an elderly population of underweight patients, who exhibited two to three times higher risk for all-cause mortality [70].

Among others, one plausible explanation for the obesity paradox in patients with AF might be the dosage of anticoagulant drugs. Numerous studies have shown that underweight patients have increased incidence of major bleeding [71]. It is interesting that in patients with AF who were under 50 kg, there seemed to be a significant increase in plasma concentrations of rivaroxaban compared with normal weight patients, whereas there was no significant difference in concentrations between the normal weight group and the obesity group [72].

Obesity Paradox and Cancer

The role of obesity in cancer has been broadly researched, highlighting that overweight and obesity have contributed to increasing the prevalence of several types of cancers [73]. Almost 55% of cancers diagnosed in women and 24% diagnosed in men are considered overweight- and obesity-related cancers, i.e., breast and endometrial cancers, adenocarcinoma of the esophagus, and multiple other sites such as the gastric cardia, colon, rectum, liver, gallbladder, pancreas, kidney, thyroid gland, and multiple myeloma [74]. The insulin/IGF1 system, the effect of sex hormones, and adipocytokines are among the mechanisms that seem to be implicated [73]. On the other hand, there are studies proposing that obesity may play a protective role in some types of cancers regarding their incidence and mortality [75]. BMI has a different effect among premenopausal and postmenopausal women, and it seems that those with obesity have a lower risk of developing pre-menopausal breast cancer, especially the hormone sensitive type, compared to lean women, a finding attributed among others to the low levels of estrogen and progesterone [76]. There are, however, other meta-analyses that have pointed out a worse breast cancer survival in both premenopausal and post-menopausal women [77, 78]. A plethora of epidemiological studies have shown a paradoxical relationship between survival and BMI among people with certain cancers. Moreover, there have been conflicting results on the effect of BMI on the prognosis of several cancers such as colorectal cancer [79], renal cancer [80], gastrointestinal stromal tumors [81], leukemia [82], B cell lymphoma [83], lung cancer [84], and esophageal cancer [85]. There are other cancer types where a positive association with BMI has been found, like the aggressive form of prostate cancer [86], while an inverse relationship has been shown with the localized type [87, 88]. There are also types of cancer for which higher BMI has been associated with poor prognosis and lower survival like postmenopausal breast cancer [77], ovarian cancers [89], and type 1 endometrial cancers [90]. Data from the Framingham cohort study showed that patients with obesity had increased mortality risk by 6–7% for every 2 years lived with obesity and the mortality from cancer for patients with obesity increased by 3% for every 2 years lived with obesity [91].

In order to understand and explore the obesity paradox in cancer, several important factors need to be considered. Essential aspects are the time of the calculated BMI, the age of the patients, their smoking habits, and the type of treatment [92]. When patients with cancer are studied, it is of great importance to acknowledge the physical status before diagnosis. When weight loss is cancer-related, it could lead to cancer cachexia characterized by unintentional weight loss. In this case, skeletal muscle mass is lost through catabolism caused directly by tumor metabolism or indirectly by other tumor-mediated effects [93]. Sarcopenic obesity has been evaluated in patients with cancers of the respiratory or gastrointestinal tract. The reported BMI had been calculated in the first day in the clinic, and several patients had presented with a trend towards weight loss in the last 6 months. They underwent a CT scan in order to quantify and calculate the fat-free mass. The results showed that 15% had sarcopenic obesity with poor prognosis and moreover sarcopenic obesity was a significant independent predictor of survival [94]. An observational study has shown that having obesity before the diagnosis of colorectal cancer is associated with increased mortality [95], whereas increased post diagnosis BMI in patients with colorectal cancer is associated with decreased mortality risk [79].

The last 50 years have strongly highlighted the large negative effect of tobacco use on health and the increased risk of development of at least 12 types of cancers [96]. The influence of tobacco use on mortality is enormous; hence, the estimation of mortality risk on patients with cancer should preferably be made separately for smokers and non-smokers [97]. The use of tobacco stimulates the sympathetic nervous system resulting in an increase of resting metabolic rate and alters the central hypothalamic regulation of appetite, resulting in reduced energy intake [98]. Therefore, tobacco users in the majority are leaner. Interestingly, several cross-sectional studies demonstrated that waist to hip ratio (WHR) is higher in tobacco users than in non-tobacco users, thus pointing towards an alteration in fat distribution [99, 100].

Other factors to be considered include the age of participants and their therapeutic treatment. A meta-analysis of 22 clinical trials examined 14 different types of cancers and treatments. The results showed that living with obesity had no influence in survival compared with normal weight patients, except in those patients treated with certain chemotherapeutic drugs accompanied with elevated mortality risk potentially due to underdosing [101]. Moreover, the age of the participants may also be relevant when the obesity paradox is under investigation. The incidence of some types of cancer might have a wide age range, e.g., leukemia. A study that examined patients with acute myeloid leukemia demonstrated that increased mortality risk was significantly associated with advanced age and low BMI [82]. The American Cancer Society in 2012 published guidelines regarding nutritional and physical activity for cancer survivors. The general recommendation for patients that have active treatment is to maintain their baseline weight and in case of overweight or obesity to mildly reduce it. The guidelines encourage patients with cancer and survivors to engage in regular physical activity and avoid inactivity. The authors underline that obesity enhances several comorbidities and preventing them is of high importance [102].

Obesity Paradox and Chronic Obstructive Pulmonary Disease

A large cohort study in 1,213,829 Koreans demonstrated that mortality risk increased as BMI decreased in patients with respiratory diseases, including COPD [103]. Likewise, a meta-analysis of 22 studies in patients with diagnosed COPD demonstrated that lower BMI was markedly associated with increased mortality risk. The highest mortality was among underweight patients, and the lower overall mortality was found in patients with overweight and obesity, even after risk adjustment analysis [104]. A study by Lainscak et al. corroborated this finding. The authors followed 968 patients with COPD for a median of 3.26 years and showed that the lowest mortality was in patients living with overweight and more specifically between the BMI range of 25.09–26.56 kg/m2 [105]. Fat-free mass was found to be an independent predictor of mortality regardless of fat mass in a cohort study of 412 patients with moderate to severe COPD. Thus, fat-free mass could be an additional tool in the assessment of COPD severity, as increased fat-free mass was associated with lower mortality [106]. Furthermore, weight changes in patients with COPD might contribute to differential health outcomes. Patients with low BMI and COPD are frequently malnourished, and they might experience significant weight loss that could even lead to cachexia [107], underlining that significant weight loss is associated with increased mortality [108]. One should keep in mind that all these studies based their quantification of the degree of obesity on BMI measurement. Information about body composition and cardiorespiratory fitness (CRF) is lacking. Low CRF is associated with high all-cause mortality and may act as a limitation bias contributing to the obesity paradox [109].

Conclusion

While the detrimental effect of obesity in mortality and morbidity has been well established, the concept of the obesity paradox in particular chronic diseases is a topic of immense interest. In certain clinical settings, obesity exhibits an intriguing “paradoxical” protective effect. The association between BMI and clinical outcomes, however, may be driven by multiple factors among which the limitations of the BMI index itself, the unintended weight loss precipitated by chronic illness, the various phenotypes of obesity, i.e., sarcopenic obesity or the athlete’s obesity phenotype, and the cardiorespiratory fitness levels of the included patients. In order to clarify the controversial issues regarding the protective role of overweight and obesity revealed in some studies, determination of BMI is not adequate, and additional indicators such as quantitative determination of different body compartments such as fat and lean mass are of paramount importance. Also, previous cardioprotective medications, obesity year duration, and smoking status seem to play an important a role, thus increasing the need for carefully designed studies alleviated by such confounding factors.