Sedentary Behavior and Incident Cancer: A Meta-Analysis of Prospective Studies

Background Sedentary behavior is ubiquitous in modern adults' daily lives and it has been suggested to be associated with incident cancer. However, the results have been inconsistent. In this study, we performed a systematic review and meta-analysis of prospective cohort studies to clarify the association between sedentary behavior and incident cancer. Method PubMed and Embase databases were searched up to March 2014. All prospective cohort studies on the association between sedentary behavior and incident cancer were included. The summary relative risks (RRs) with 95% confidence intervals (CIs) were estimated using random effect model. Results A total of 17 prospective studies from 14 articles, including a total of 857,581 participants and 18,553 cases, were included in the analysis for sedentary behavior and risk of incident cancer. The overall meta-analysis suggested that sedentary behavior increased risk of cancer (RR = 1.20, 95%CI = 1.12–1.28), with no evidence of heterogeneity between studies (I 2 = 7.3%, P = 0.368). Subgroup analyses demonstrated that there were statistical associations between sedentary behavior and some cancer types (endometrial cancer: RR = 1.28, 95% CI = 1.08–1.53; colorectal cancer: RR = 1.30, 95%CI = 1.12–1.49; breast cancer: RR = 1.17, 95%CI = 1.03–1.33; lung cancer: RR = 1.27, 95%CI = 1.06–1.52). However, there was no association of sedentary behavior with ovarian cancer (RR = 1.26, 95%CI = 0.87–1.82), renal cell carcinoma (RR = 1.11, 95%CI = 0.87–1.41) or non-Hodgkin lymphoid neoplasms (RR = 1.09, 95%CI = 0.82–1.43). Conclusion The present meta-analysis suggested that prolonged sedentary behavior was independently associated with an increased risk of incident endometrial, colorectal, breast, and lung cancers, but not with ovarian cancer, renal cell carcinoma or non-Hodgkin lymphoid neoplasms.


Introduction
Sedentary behavior is ubiquitous in modern adults' daily lives [1]. It is defined as any waking behavior in a sitting or reclining posture, expending #1.5 times the resting energy demand (for example TV viewing, computer use, occupational sitting, reading, and sitting in a car) [2]. Sedentary behavior is distinct from physical inactivity (i.e. not meeting sufficient levels of moderate-tovigorous physical activity (MVPA)) [2]. The time adults spend sedentary is relatively independent from their time spent in MVPA, for example, individuals may frequently participate in MVPA but still spend substantial amounts of their time sitting [3]. Estimates derived from objective accelerometry suggest that adults spend about 50-60% of their waking day sitting [4]. TV viewing in particular is one of the most prevalent sedentary behaviors, occupying 40% of daily leisure time in some European countries and about 50% in Australia and in the USA [5]. Sedentary behavior may have a detrimental effect on health outcomes, as shown by recent meta-analyses providing evidence that prolonged sedentary behavior increases the risk of the metabolic syndrome [6], type 2 diabetes, cardiovascular disease, and all-cause mortality [7][8][9].
In this study, we restricted ourselves to reviewing prospective cohort studies since cross-sectional or case-control studies are subject to recall bias and even reverse causality. Thus, we performed a systematic review and meta-analysis of all published prospective studies to further clarify the association between sedentary behavior and incident cancer.

Inclusion criteria and data extraction
Studies included in the meta-analysis met all the following inclusion criteria: (1) evaluated the association between sedentary behavior (total sitting time, occupational sitting time, leisure sitting time or TV viewing) and incident cancer; (2) used a prospective cohort design; (3) provided relative risks (RRs) or hazard ratios (HRs) with 95%CIs for highest versus lowest level of sedentary behavior. The following information was extracted from each study: (1) the first author; (2) publication year; (3) country name; (4) sex distribution; (5) age distribution of study population at baseline; (6) average duration of follow-up; (7) number of cases and study population; (8) types of cancer; (9) RR or HR with 95%CI for highest versus lowest level of sedentary behavior; (10) covariates used in adjustment. Two authors independently assessed the articles for compliance with the inclusion/exclusion criteria and resolved disagreements through discussion.

Statistical analysis
Random [25] effects model was used to calculate pooled RRs with 95%CIs for the highest versus the lowest level of sedentary behavior. Sensitivity analysis, after exclusion one study at each time, was applied to test the stability of the results. Subgroup analyses were performed to investigate the association between sedentary behavior and risk of types of cancer. In addition, we also tested the association between TV viewing and risk of cancer. Publication bias was assessed by Begg's test [26] and Egger's test [27] (P,0.05 was considered statistically significant). Statistical

Characteristics of included prospective studies
Following the literature search and selection, a total of 17 prospective studies from 14 publications were included in the meta-analysis examining the association between sedentary behavior and risk of incident cancer (Figure 1). The duration of follow-up ranged from 3.8 to 16 years. 13 studies originated from America, 2 study from Europe, and 2 studies from East Asia. The characteristics of included prospective studies are listed in Tables 1  and 2.

Sedentary behavior and incident cancer
17 prospective studies from 14 articles were included in the meta-analysis, including a total of 857,581 participants and 18,553 cases. The overall meta-analysis suggested that sedentary behavior increased risk of cancer (RR = 1.20, 95%CI = 1.12-1.28), with no evidence of heterogeneity between studies (I 2 = 7.3%, P = 0.368) ( Figure 2). Sensitivity analysis suggested that the result was stable, with ORs and 95%CIs ranging from 1.18 (1.11-1.26) to 1.22 (1.15-1.31). There was no publication bias (P = 0.202). It should be noted that RRs and 95% CIs with adjustment for potential confounding factors, such as BMI, physical activity and energy intake, from all included studies were pooled together using metaanalysis.
In the subgroup analyses (Figure 3), there were statistical associations between sedentary behavior and some cancer types (endometrial cancer: RR = 1.28, 95% CI = 1.08-1.53; colorectal cancer: RR = 1. Since TV viewing is the main type of sedentary behavior, we also investigated the association between TV viewing and risk of cancer. The results suggested that sedentary behavior increased the risk of cancer (RR = 1.21, 95%CI = 1.08-1.35) (Figure 4). We did not performed subgroup analyses based on type of cancer again because of limited studies for each cancer type.

Discussion
Our meta-analyses suggest that prolonged sedentary behavior is associated with an increased risk of some cancer types including colorectal, lung, breast, and endometrial cancers, but not with ovarian cancer, renal cell carcinoma or non-Hodgkin lymphoid neoplasms. The positive association was independent of traditional risk factors including BMI, physical activity and energy intake.
In many Western countries, adults spend large proportions of their awake time sedentary. It is estimated that the average US population spends about 35 h/week watching TV, 2 h/week watching time-shifted TV, and 4 h/week on the internet [28]. Although insufficient physical activity has long been considered as a risk factor of many chronic diseases (e.g., type 2 diabetes [29], coronary heart disease [30], cancer [31]) and all-cause mortality [32], examining the independent relationships between sedentary time and health outcomes is fairly recent. In 2011, a meta-analysis performed by Grøntved and Hu [7] reported that each 2-hour increment of TV viewing daily was associated with increased risk of type 2 diabetes (RR 1.20, 95%CI = 1.14-1. 27 [33] reported that sedentary behavior was associated with an increased risk of colon cancer. However, their conclusion is mainly based on case-control studies and also they did not examine the association between sedentary behavior and risk of other types of cancer.
Several plausible mechanisms may explain the observed association between sedentary behavior and risk of some cancers. First, when a substantial amount of time is spent sitting, especially in front of the TV, this automatically means less physical activity, and it may lead to a higher energy intake [34], both which affect energy balance in such a way that possibly overweight/obesity results. However, the observed associations remained when adjusting for BMI or WC, energy intake, and physical activity. In addition, BMI or WC may then act more like a confounder. As is known, obese individuals are more prone to stay sedentary than non-obese ones. Second, excessive sitting time could increase levels of inflammatory factors such as tumor necrosis factor-a, interleukin-6, and leptin which are known risk predictors for cancer [35]. Third, sitting time has been hypothesized to influence sex hormones which could affect immune function [24].
Strengths of our meta-analysis include the prospective design, large sample size, long duration of follow-up, and ability to control for many potential confounding factors for the included studies. However, several limitations should be considered. First, sedentary behavior time in all included studies was self-reported, which could have resulted in the possibility of exposure misclassification and thereby underestimation of the true association between sedentary behavior and risk of cancer. Second, sedentary behavior was only measured at baseline for most included studies, and individuals may have changed their sedentary lifestyle during follow-up. This misclassification may also attenuate the true association. Third, although many potential confounding factors have been adjusted for, residual confounding because of poorly measured or unmeasured confounding factors may influence our results, particularly also for MVPA, a key confounding variable [36]. In addition, we can not rule out residual confounding of smoking on lung cancer. Fourth, we were unable to distinguish between most types of sitting (i.e., TV viewing, reading, using computer, sitting at work) which may have different levels of energy expenditure or be associated with different underlying confounding structures. Fifth, there was limited number of studies for each cancer site. However, the total number of cancer cases for each cancer site was relatively large (all n.1000 for each cancer site). Sixth, the included 17 studies adjusted for different confounding factors, which might have influenced the results.
In this study, we found significant associations between sedentary behavior and risk of some types of cancer including colorectal, lung, breast, and endometrial cancers. Public health guidelines for prevention and control of incident cancer may need to consider recommendations about reducing time spent sitting in addition to increasing MVPA. Actually, World Cancer Research Fund report has already made a statement on the importance of limiting sedentary behavior. Our current findings based on prospective studies further supported this statement.