1. Introduction
Colorectal cancer (CRC) is the third most commonly diagnosed cancer in Europe, with an estimated 499,667 new cases in 2018, and is the second most common cause of cancer mortality [
1]. Approximately 3–5% of all CRC cases may be due to hereditary cancer syndrome, also known as Lynch syndrome (LS) [
2], which is caused by pathogenic germline variants in DNA mismatch repair genes (
path_MMR):
MLH1,
MSH2,
MSH6, or
PMS2 [
3]. Individuals with
path_MMR variants are at considerably greater risk of developing CRC (40–80%), endometrial cancer (40–60%) and various other cancers compared to the general population [
4,
5]. However, cancer risk is highly variable among different mutation carriers and thus far it is not known why not all
path_MMR carriers develop cancer, whereas others develop cancer at a young age and/or suffer from multiple different types of cancers during their lifespan.
There is strong evidence derived from the general population that increased physical activity and reduced body adiposity are associated with decreased cancer risk [
6,
7,
8,
9,
10,
11,
12]. Recent research suggests that LS CRC risk can also be moderated by these lifestyle factors [
13,
14,
15,
16,
17], but the number of studies that investigate the associations between lifestyle and LS cancer risk are scarce. Currently, there are only two studies that we are aware of that have assessed the association of physical activity and LS cancer risk [
13,
14] and they have not taken into consideration the fact that this potential association may vary during a lifespan. In addition, the evidence regarding the associations of cancer risk and obesity and overweight is contradictory among women with LS [
15], and thus, the risk analyses should be performed separately for both sexes. It is of great importance to identify modifiable behavioural factors of LS cancer risk to motivate variant carriers to change suboptimal conduct or maintain a healthy lifestyle. Lifestyle modification could efficiently aid in reducing individual cancer risk despite a strong genetic predisposition.
In this study, we hypothesized that the interplay between genetic factors and lifestyle is associated with variable cancer risk in a distinct high-risk population. Two founder mutations in
MLH1 are found in a major proportion of Finnish LS families [
18,
19,
20], which offers a possibility for investigating lifestyle factors as a modifier of cancer risk in a relatively homogenous LS population and hence may limit the influence of genetic discrepancies. The aim of this retrospective study with longitudinal lifestyle and cancer register data was to investigate associations between body weight and physical activity on CRC and overall cancer risk among adult Finnish
path_MMR men and women. We modeled the recalled levels of weight and physical activity as time-dependent variables in the relative risk model.
3. Discussion
We conducted a retrospective study with longitudinal data collection and cancer register data among Finnish path_MMR carriers to elucidate the associations between changes in adult body weight, physical activity, and cancer risk. Our results suggest that associations between lifestyle and cancer risk differ between men and women and may vary during the course of life. We found that an overall increase in total body weight throughout the lifespan slightly elevated the risk of cancers, including CRC, in men. We also observed that men who continued to participate in more intensive physical activities over their adult life were at lower risk of all cancers.
In Western societies, body weight typically accumulates during the adult lifespan and growing levels of obesity predispose individuals to multiple health complications. Obesity is acknowledged as one of the most important risk factors of non-communicable diseases [
9,
21,
22]. It is well established in the general population that excess body weight and adiposity, particularly in overweight individuals, is an important risk factor for several cancers [
8] and the risk could be reduced by lowering excess body mass [
23]. In the population of
path_MMR carriers of the current study, participants increased their body weight during their adult years, and 61% of men and 55% of women were overweight or obese. Obesity and overweight may be more harmful for men, as we found that a trend of body weight accumulation during adult years was associated with an increased cancer risk in men but not in women. This is in agreement with other reports which have investigated weight accumulation in relation to the risk of LS cancer [
17,
24].
Unfortunately, in most of the studies, including ours, a lifelong change in body composition could only be determined by changes in body weight. However, aging is not only associated with an increase in body weight that is related to fat accumulation but also to changes in body composition [
25]. Beginning from the age of 30 years, muscle mass tends to decrease and the decline accelerates after the age of 50 years, particularly among women due to menopause [
26]. Concurrently, the amount and distribution of body fat may change, thereby resulting in the accumulation of fat—particularly in visceral areas—which increases cancer risk through several already identified biological pathways. The best-characterized association is between abdominal obesity and disturbed insulin metabolism, which may influence cancer risk through cell proliferation and apoptosis [
10,
27]. Age-related trends in body weight accumulation are different between men and women, which may explain our dissimilar findings regarding weight accumulation and cancer risk. In general, men are more prone to increased android-type fat distribution—that is, abdominal fat distribution—throughout their lifespan [
28]. In contrast, women tend to be more prone to gynoid-type fat distribution during their premenopausal years and then shift to androgen-type fat accumulation in their postmenopausal years [
29]. In our study, the weight accumulation was lower among women, but we cannot exclude the potential confounding role of menopause on the association between body weight and cancer risk as our data did not include information regarding the menopausal status of the female participants. Based on the population averages of menopause age being between 50 years and 53 years, we can estimate that 67% of women participating in this study were post-menopausal, but the considerable individual variation observed makes this estimate imprecise [
30]. Nevertheless, visceral adiposity has been associated with higher cancer risk in both sexes, although in women the risk estimates appear to differ between pre- and post-menopausal women [
31].
Intriguingly, we also found that near-term weight increase had a CRC-protective effect in women. To speculate, it is possible that hormonal factors might have influenced the risk estimates. The primary source of endogenous estrogen in post-menopausal women has been suggested to be adipose tissue [
32,
33]. Therefore, higher adiposity could maintain a higher systemic estrogen level, which in turn may provide some protection through (for instance) the anti-inflammatory action of estrogens [
34]. However, this is highly speculative. There is evidence that exogenous estrogen use can be cancer protective, neutral or to increase the risk of different cancers [
35,
36,
37]. In the current study, we did not investigate the potential role of hormone therapy, nor did we have the ability to measure estrogen levels; thus, we cannot exclude the role of systemic estrogen level. Therefore, as the great majority of CRC-diagnosed women in our study were over 60 years of age, there remains a possibility that the plausible protective effects of estrogen derived from adipose tissue might have masked the effect of weight gain on colorectal cancer risk.
Our results suggest that performing more vigorous guided physical activity exhibit a cancer-preventive effect in men, as those who continued at higher levels of physical activity were at a 63% lower cancer risk when compared to less active men performing non-guided physical activity. To date, we are aware of only two previous studies that have assessed the impact of physical activity on LS cancer risk [
13,
14]. Both designs were retrospective like our study but did not assess risk estimates separately for men and women. Kamiza et al. (2015) [
13] reported that among 301 Taiwanese individuals (51.8% women) carrying
path_MLH1 and
path_MSH2 regular vigorous leisure time physical activity over a year prior to cancer diagnosis decreased CRC risk by 38% when compared to those who did not indulge in any such activity. Although we did not observe associations between physical activity and cancer risk in the near-term like Kamiza et al. (2015) [
13], similarly to their findings, our results also suggested that performing vigorous physical activity could reduce cancer risk.
Further, the study by Dashti et al. (2018) [
14] comprised 2042
path_MMR carriers (57% women). As in our study, they modeled longitudinal and near-term changes in physical activity separately. Unlike our study, Dashti et al. (2018) [
14] did not find an association between physical activity and cancer risk when assessed over several age periods, even though a trend of lowering the risk of CRC was observed. In addition, they used MET-h/week to assess the amount of physical activity, but we did not do so. Near-term cancer risk assessment, which was used in the current study, may be particularly effective for identifying specific suboptimal lifestyle changes that could be associated with carcinogenesis and precede cancer occurrence. Dashti et al. (2018) [
14] found higher levels of near-term physical activity (>35 MET-h/week) to be protective against such a risk, whereas we did not find an association between near-term physical activity and cancer risk (any cancer or CRC).
We did not find longitudinal or near-term evidence linking physical activity to cancer risk in
path_MMR women, which may be due to differences in physical activity behavior between men and women, including the intensity and type of physical activity, as well as the timing of physical activity exposure in life. For example, throughout the follow-up, female participants mainly reported participating more frequently in guided leisure time physical activity than men, who reported performing competitive sports more often (
Table S4). Overall, our results highlight the potential role of physical activity in cancer prevention among
path_MMR carrier men, as already advocated in clinical guidelines for LS [
38]. Although most
path_MMR carriers suffer from cancer at some point of their life, this is an important finding.
Various reports in the extant literature have suggested several mechanisms that link physical activity with a reduced cancer risk [
11]. For example, physical activity produces multiple beneficial changes in cardiorespiratory systems [
39], and being physically active also helps with weight control, as well as with reducing excess adiposity [
12]. These combined effects might have a beneficial impact on biological mechanisms that interact directly or indirectly with cancer, such as improved insulin sensitivity and reduced chronic low-level inflammation, which is also linked with favorable immunomodulation [
40,
41]. However, the existing evidence originates from sporadic cancer patients who could differ from
path_MMR carriers with respect to disease mechanisms, carcinogenesis, and biological regulation.
As described previously, the association of decreased cancer risk and healthy lifestyle in the general population has been observed in large population-based studies. Since the influence of healthy behaviour on decreased cancer risk was observed in our study with a limited number of participants, it could be possible that the effect of the modifiable behavioral risk factors—physical activity and body weight—is emphasised in
path_MMR carriers due to their strong genetic predisposition to cancers. Therefore, it is important to follow these modifiable risk factors among
path_MMR carriers during their regular healthcare visits. An optimal lifestyle could partially compensate for the strong genetic predisposition to cancers and thus help in cancer prevention. Nowadays, individual cancer risk can be calculated and demonstrated via online tools (
www.plsd.eu, accessed 9 January 2021), which can be used to improve motivational support for healthy lifestyle maintenance or lifestyle changes.
A major strength of the current study is that the study cohort comprised participants who had undergone comprehensive screenings of LS-predisposing mutations, with ascertainment utilizing Amsterdam and Bethesda clinical criteria and cascade testing, and those who had been offered colonoscopy surveillance at 2–3-year intervals. Our body weight and physical activity data collection encompassed the entire adult lifespan and was carefully analyzed by considering potential time-varying risks, sex differences, and potential confounders regarding cancer incidence. The observation period was initiated from the age of 20 years, instead of birth, to avoid the detection of changes in body weight which were merely due to natural growth and maturation. In doing so, we were also able to exclude the time-period when cancer incidence tends to be extremely low even among path_MMR carriers. We also chose to model the change in cancer risk in the near-term setting (during the age-period of cancer or censoring) as it could be more accurate than longitudinal change, which could be influenced by poor recall. We also used time-dependent covariate values, which allow us to account for changes in predictor values over time.
However, there are also several limitations. Weight and physical activity were assessed using self-recall instruments. Even though a recent study found that cross-sectional self-reported measurements of BMI were reasonably close to recent direct measurements [
42], the recall of weight in the more distant past has lower reliability [
43] and for some of our older participants, the recall time was several decades. Moreover, there might also be sex-based discrepancies, as women tend to underestimate their weight and men tend to overestimate it [
44]. Finally, Smith et al. (2013) [
45] found the recall of physical activity of the distant past to be moderately reproducible, but poor at the individual level. Taken together, we cannot exclude the possibility that recall bias might have influenced the risk estimates.