1. Introduction
Breast cancer is the most commonly diagnosed female cancer and endometrial as well as ovarian cancer incidence rank sixth and eighth, respectively, underscoring the global impact caused by these cancers [
1]. Several meta-analyses and epidemiological studies implicated obesity as a player in the development of female-specific cancers [
2,
3,
4]. In observational studies associations between body mass index (BMI) as a measure of general obesity and an increased risk of postmenopausal breast [
5] and a decreasing risk of premenopausal breast cancer [
6] were found. Obesity is the most established risk factor for endometrial cancer [
7,
8] and may be positively associated with ovarian cancer [
9,
10]. In recent Mendelian randomization (MR) analyses genetically predicted BMI was positively associated with endometrial cancer [
11,
12], but (contrary to the findings of observational studies) inversely associated with both pre- and postmenopausal breast cancer [
13,
14,
15]. Another MR study reported that genetically predicted BMI was positively associated with non-high grade serous ovarian cancers but was unrelated to the more common and aggressive high-grade serous ovarian cancer [
16].
A growing body of evidence suggests that body composition may play an important role in site-specific cancer development [
17,
18,
19], but the data are inconclusive. Some investigations focusing on abdominal fat distribution as represented by the waist-to-hip ratio (WHR) found a positive association with breast cancer [
20,
21]. In a recent prospective observational study waist circumference (WC) and BMI were positively associated with endometrial and premenopausal breast cancers, but not with ovarian cancer [
22]. Whether abdominal fat distribution is associated with the female-specific cancers independently of BMI remains to be clarified.
The present study is the first MR study using bioelectrical impedance analysis measurements of body fat distribution in addition to BMI as potential risk factors for female-specific cancers. We sought to generate more comprehensive and robust evidence of the impact of trunk, arm, and leg fat ratios (TFR, AFR, LFR) on breast, ovarian, as well as endometrial cancer including their histotypes based on univariable and multivariable MR analyses.
4. Discussion
Our MR study showed that visceral adiposity increased the risk of breast cancer, mainly ER+ breast cancer, and the risk of ovarian cancer, mainly clear cell ovarian cancer. Relations of genetically predicted visceral adiposity to breast and endometrial cancers were weaker than those with general adiposity. However, visceral adiposity appears to play a more important role than general adiposity for the risk of ovarian cancer, in particular the clear cell and endometrioid cancer subtypes. The findings confirmed a causal protective effect of general adiposity on breast cancer risk. This effect could be ascribed to associations with both ER- and ER+ breast cancers and was confirmed in multivariable analyses. In contrast, general adiposity was a strong causal risk factor for endometrial cancer and a weaker causal risk factor for ovarian cancer, in particular for the endometrioid histotype. LFR was not related to overall cancers but showed effects in the same direction as AFR on ER+ breast cancer and clear cell ovarian cancer.
Epidemiologic studies reported a positive relationship between obesity and postmenopausal breast cancer and an inverse association with premenopausal breast cancer [
30,
31,
32,
33]. One study using data from two chemoprevention trials found a positive association in premenopausal women [
34]. In contrast, two prior MR studies showed that adult BMI is inversely related to postmenopausal breast cancer [
13,
14]. We also found an inverse causal effect of overall fat content on breast cancer risk; however, we were unable to distinguish between pre- and postmenopausal breast cancers. The apparent adverse effect of adult body size on breast cancer risk may be attributable to a large body size in childhood that persists into adulthood [
35]. Of note, childhood BMI showed inverse associations with both premenopausal and postmenopausal breast cancer [
36,
37]. Because adult weight gain is a strong predictor of postmenopausal breast cancer risk [
38], the positive association between adult BMI and postmenopausal breast cancer shown in epidemiologic studies is likely due to the effect of weight gain in adulthood. Weight gain may better reflect the dynamics of changes in body shape over time, with increased visceral and metabolically active fat accumulation, especially during menopause [
39]. According to the WCRF, abdominal obesity and weight gain in adulthood convincingly increases the risk of postmenopausal breast cancer [
31]. Our finding that genetically predicted TFR showed a positive effect on ER+ breast cancer suggests that body fat distribution plays a particular role in postmenopausal breast cancer. Supporting the assessment of the IARC Working Group, [
30], we showed that an increasing BMI and AFR had a strongly positive effect on endometrial cancer. However, we found no evidence for a causal link between abdominal adiposity and endometrial cancer; this is in contrast to umbrella reviews [
3,
4,
31], in which the association between WC and WHR and total endometrial cancer was supported by strong evidence. Our results support the findings of recent MR studies, where genetically elevated BMI, but not WHR, was found to be causally related to endometrial cancer risk [
11,
40].
The evidence relating body fatness to ovarian cancer and its subtypes is inconclusive [
41]. The WCRF CUP graded the relationship as probably causal [
31] and the IARC suggested there is strong evidence for a positive association between obesity and ovarian cancer [
30]. Likewise, a systematic review and meta-analysis reported an increased risk of ovarian cancer with higher BMI, and a marginally significant positive association with WC, but no association with hip circumference or WHR [
42]. An umbrella review of dose-response meta-analyses rated the evidence for an association between BMI and ovarian cancer as weak [
43], but another umbrella review [
2] rated the evidence as suggestive only [
2]. Using a MR approach, Gao et al. [
14] found a positive association between BMI and ovarian cancer, which is in line with our results; contrary, a one-sample MR study found no association between both BMI and fat mass and ovarian cancer [
44]. However, our study was two-sampled, included more cases, and used larger GWAS with 201 more genetic instruments for the analyses resulting in higher statistical power. We further showed a positive association between TFR and ovarian cancer, in particular for the endometrioid and clear cell histotypes, a result that extends the existing evidence and may contribute to better understanding of the pathogenesis of this cancer.
Obese adipose tissue seems to create a pro-oncogenic environment. Specific biological mechanisms to explore the link between obesity and cancer risk are centered along metabolic and endocrine consequences of body fat accumulation. Changes in the metabolism of sex hormones, insulin and insulin-like growth factor signaling, inflammatory effects and Adipokine-related effects were identified. According to an IARC working group, the evidence for a role of sex hormones and chronic inflammation was judged as strong, while the evidence for insulin and IGF was moderate [
30]. Among other factors, obesity-related changes in the tumor microenvironment, cellular perturbations, and the intestinal microbiome are likely to affect cancer development [
45]. In addition, obesity-related changes in anti-tumor immunity and angiogenesis were observed [
46]. Biological mechanisms that could explain the inverse effects of BMI and AFR on the development of breast cancer and its subtypes are widely unknown.
Our study has a number of considerable strengths. Using both BMI and AFR we could replicate the findings of a previous MR study, which provided support for our estimates. While BMI approximates total fat content, the fat ratio indexes used in our study discriminated between adipose and lean mass and are therefore more reflective of actual fat mass and body fat distribution. All exposures were modelled as continuous variables to avoid residual confounding and to boost statistical power. Sufficiently large F-statistics reduced the potential for weak instrument bias. The broad range of robust methods allowed us to strengthen our findings by considering different patterns of pleiotropy. The multivariable approach allowed us to distinguish between the direct effects of overall and abdominal fat content.
Our study also has certain limitations. TFR considers the fat content of the entire torso rather than that of the abdomen only, which may have underestimated causal associations. MR analyses assume relationships to be linear. Quantitative estimates may be misleading if the true relationship is non-linear; although estimates are still reflective of the presence and direction of the population-averaged causal effect [
47]. Our analyses were based on European women and thus, results cannot be transferred to other ethnicities. Finally, we were unable to distinguish between pre- and postmenopausal breast cancer.