IJC International Journal of Cancer

Transcription

1 IJC International Journal of Cancer Association between adult weight gain and colorectal cancer: A dose response meta-analysis of observational studies Qi Chen 1, Jing Wang 1, Jinghui Yang 2, Zhichao Jin 1, Wentao Shi 1, Yingyi Qin 1, Feifei Yu 1 and Jia He 1 1 Department of Health Statistics, Second Military Medical University, Shanghai, China 2 Department of Organ Transplantation, Changzheng Hospital, Shanghai, China This study investigated the association between adult weight gain and risk of colorectal cancer (CRC). Using terms related to weight gain and CRC, we searched PubMed, Embase and Web of Science for relevant studies published before June Two evaluators independently selected studies according to the selection criteria, and eight studies were included (three case control and five cohort studies). Summary estimates were obtained using fixed- or random-effects models. The relative risk (RR) of the association between adult weight gain and CRC was 1.25 (95% confidence interval [CI], ); the RR was 1.30 (95% CI, ) for colon cancer (CC) and 1.27 (95% CI, ) for rectal cancer (RC) for the highest versus lowest category. For every 5-kg increase in adult weight, the risk increased by 5% (RR, 1.05; 95% CI, ) for CRC, 6% (RR, 1.06; 95% CI, ) for CC and 6% (RR, 1.06; 95% CI, ) for RC. The subgroup analyses showed a positive association between adult weight gain and risk of CRC only in men, and the RR was 1.65 (95% CI, ) for the highest versus lowest category of adult weight gain and 1.10 (95% CI, ) for a 5-kg increase in adult weight. In conclusion, there is evidence that adult weight gain is associated with an increased risk of CRC. However, the positive association between adult weight gain and risk of CRC is stronger among men than among women. Colorectal cancer (CRC), which includes colon cancer (CC) and rectal cancer (RC), is the third most common cancer in men and the second most common in women worldwide 1 ;it is responsible for a significant proportion of the global burden of cancer morbidity and mortality. 2 In recent decades, there has been a swift increase in CRC morbidity in rapidly developing countries where lifestyles have changed significantly. 3 Many studies have demonstrated that some lifestyle factors may contribute to CRC, 4 6 specifically obesity. 7 The increasing worldwide prevalence of overweight and obesity 8 may lead to a higher incidence of CRC. Although excess body weight as measured by body mass index (BMI) and waist circumference (WC) have been considered as risk factors for CRC, 9 the magnitude of the association has varied widely by weight assessment period. 10 Adult Key words: adult weight gain, colorectal cancer, meta-analysis Q.C., J.W., J.Y. and Z.J. contributed equally to this work. Grant sponsor: Leading Talents of Science in Shanghai; Grant number: 2010(022); Grant sponsor: Key Discipline Construction of Evidence-Based Public Health in Shanghai; Grant number: 12GWZX0602; Grant sponsor: Key Program of Shanghai Soft Science Research; Grant number: DOI: /ijc History: Received 15 July 2014; Accepted 3 Nov 2014; Online 14 Nov 2014 Correspondence to: Jia He, Department of Health Statistics, Second Military Medical University, No. 800 Xiangyin Road, Shanghai , China, Tel.: [ ], Fax: [ ], hejia63@yeah.net weight gain is a dynamic measure reflecting fluctuations in weight over time, which is thought to be a better measure of adiposity than a static measure of weight such as BMI. 11 Also, cancer has a complex etiology, and the process of tumor initiation, promotion and progression can take decades. 12 Adult weight gain during early to middle adulthood has an important influence on the risk of CRC. 13 Several studies investigating the association between adult weight gain and risk of CRC have yielded inconsistent findings. Some studies have demonstrated a significant correlation, but other studies have failed to show any significant correlation. 12,19,20 This lack of agreement between results of previous studies may be due to differences in cancer site, sex, study population or residual confounding. Because of the importance of this possible association in both clinical practice and public health, we systemically reviewed previous studies to explore the association between adult weight gain and risk of CRC. Also, we conducted a dose response meta-analysis to quantitatively assess the dose response relationship between adult weight gain and risk of CRC. Material and Methods Data sources and search strategy Observational studies (case control, nested case control and cohort studies) on adult weight gain and risk of CRC were included in our meta-analysis, not taking into account language, publication status or article type. Two investigators (Q.C. and J.W.) conducted a systematic literature search by using the electronic databases PubMed (from 1965 to June

2 Chen et al What s new? It s tricky to study the relationship between weight gain and colorectal cancer, and previous studies have returned conflicting results. This meta-analysis collated data from 8 studies in search of a clear indication of the effect of adult weight gain on CRC risk. Unlike body mass index, which includes both fat and muscle mass, adult weight gain can reveal changes in metabolic efficiency with age that make one vulnerable to cancer. These authors found that only in men, weight gain does appear to increase risk of colorectal cancer. 2014), Embase (from 1965 to June 2014) and Web of Science (from 1986 to June 2014). Searches were performed using Medical Subject Heading terms and the free keywords colorectal cancer, colorectal neoplasm, bowel cancer, colon cancer, colon neoplasm, rectal cancer and rectal neoplasm combined with weight change and weight gain. Additionally, all of the references cited in relevant original and review articles were searched manually. Selection criteria Eligible studies met the following inclusion criteria: (i) it was an observational study (case control study, nested case control study or cohort study); (ii) the exposure of interest was adult weight gain; (iii) the outcome of interest was CRC, CC or RC; (iv) it reported the relative risk (RR) or odds ratio (OR) and the 95% confidence interval (CI) for the association between adult weight gain and risk of CRC; and (v) adult weight gain was calculated as weight at baseline age minus weight at early adulthood (approximately 20 years of age). Studies that did not document the incidence of CRC, animal experimentation studies and mechanistic research studies were excluded. Also, to avoid combining studies that were not comparable, studies for which the exposure of interest was annual adult weight gain or adult BMI gain were excluded. Study selection was conducted independently by two investigators (Q.C. and J.W.). Data extraction and quality assessment Data extraction and quality assessment were independently performed by two reviewers (Q.C. and J.Y.). The following information was extracted from each eligible study: first author s surname, year of publication, study location, study design, source of study population, sample size, number of events, proportion of male subjects, reference age, definition of weight change, type of cancer, estimated effect size (OR or RR), 95% CI and covariates adjusted in the statistical analysis. In addition, we extracted the number of cases, number of non-cases or person-years, effect size (OR or RR) and 95% CI of individual weight gain categories for dose response meta-analysis. In cases of studies that reported several multivariable-adjusted effect estimates, the result that was fully adjusted for potential confounding variables was selected. Quality assessment was conducted by using the 9-star Newcastle Ottawa Scale (NOS). 21 We considered studies with an NOS score of 7 or more to be high-quality studies. After data extraction and assessment, information was examined and adjudicated independently by an investigator (J.H.) who referred to the original articles. Statistical analysis We examined the relationship between adult weight gain and risk of CRC on the basis of the RR and 95% CI published in each study. Because the absolute risk of CRC is low, the OR in case control studies approximate the RR. 22 We first performed a meta-analysis comparing the risk of CRC between the highest and lowest category of adult weight gain in all included studies. CRC was defined as colon and rectal cancer, and colon or rectal cancer if the original articles presented results of colon or rectal cancer separately. 9 If a single study reported results for different populations (male and female) but did not report the overall results, the results for each population were calculated as a different study. 2 Subsequently, we conducted a dose response meta-analysis from the correlated natural log of RRs across categories of adult weight gain, 23,24 to estimate linear trends of RR for CRC per 5-kg increment in weight gain as two included studies. 15,17 A fixed-effect model was used to estimate the pooled RR with 95% CI if there was no evidence of heterogeneity; otherwise, a random-effect model was used. 25 Forest plots were used to visually assess the RR estimates and corresponding 95% CI. We performed a two-stage random-effect dose response meta-analysis to examine the potential nonlinear relationship between adult weight gain and risk of CRC. This was conducted by modeling adult weight gain using restricted cubic splines with three knots at fixed percentiles (10, 50, and 90%) of the distribution. The details of the methods were described by Larsson and Orsini. 26,27 The p-value for linearity was calculated by testing the null hypothesis that the coefficient of the second spline was equal to zero. This method requires that the distributions of cases and person-years or non-cases and the RRs with the variance estimates for at least three quantitative exposure categories are known. However, one study did not report the distributions of non-cases for each exposure category but only disclosed the total number of participants. 13 We estimated the distribution of non-cases for each category in this study using methods described by Aune et al. 28 We assigned the dose of adult weight gain from each study to these categories based on the calculated midpoint of adult weight gain if the study did not report the median of

3 2882 Adult weight gain and colorectal cancer summary RR was 1.25 (95% CI, ) for CRC with some heterogeneity (I %, p ) and no publication bias (Egger test p ), 1.30 (95% CI, ) for CC with some heterogeneity (I %, p ) and no publication bias (Egger test p ) and 1.27 (95% CI, ) for RC with small heterogeneity (I %, p ) and no publication bias (Egger test p ). As shown in Figure 3, there was a linear relationship between adult weight gain and risk of CRC (p ), CC (p ) and RC (p ). Table 2 shows the summary RRs of a 5-kg increment of adult weight gain, which were 1.05 (95% CI, ) for CRC, 1.06 (95% CI, ) for CC and 1.06 (95% CI, ) for RC. Also, the pooled RR of CRC, CC and RC calculated from the spline model for 5, 10, 15 and 20 kg of adult weight gain compared with that for no adult weight gain is shown in Table 2. Figure 1. Selection of studies for inclusion in meta-analysis. the weight gain category. If the upper or lower boundary of the highest or lowest category of adult weight gain was not provided, we assumed that it had the same amplitude as the closest category. 29 Heterogeneity between studies was evaluated by using the v 2 test and I 2 statistic. 30 The probability of publication bias was assessed by using the Egger regression test. 31 If publication bias existed, we evaluated the effect of publication bias by using the trim and fill method. 32 Subgroup analyses were performed according to different cancer sites (CC and RC), geographic areas (North America and other areas) and sex (male and female). Sensitivity study analyses were performed after excluding retrospective case control studies. Because all included studies were well conducted (NOS score 7), we did not perform sensitivity analyses according to NOS score. Stata Version 12.0 software (Stata Corp., College Station, TX) was used for all analyses, and all statistical tests were two-sided. A p-value < 0.05 was considered to be statistically significant. Results Description of the selected studies In June 2014, 594 records were retrieved by using our search strategy. After reading the titles and abstracts, we excluded 551 records and retained 43 studies for further evaluation by reading the full text. Finally, we identified eight full-text studies on adult weight gain and CRC for our meta-analysis, including five cohort studies 12,14,15,19,20 and three case-control studies The search process is depicted in Figure 1. These eight studies included 394,434 participants and 11,091 cancer patients. The general characteristics of the included studies are presented in Table 1. Adult weight gain and risk of CRC Figure 2 shows the summary RR of CRC, CC and RC for the highest versus lowest categories of adult weight gain. The Sensitivity analysis and subgroup analysis The results of sensitivity analyses and subgroup analyses are listed in Table 2. When the analysis was restricted to cohort studies only, the summary RR of the highest versus lowest category and of an increment of 5 kg of adult weight gain was 1.15 (95% CI, ) and 1.04 (95% CI, ), respectively, for CRC risk. Stratifying by geographic region, the RR was 1.21 (95% CI, ) and 1.36 (95% CI, ) for CRC in North America and other areas, respectively, when comparing the highest versus lowest categories of adult weight gain. The RR of a 5-kg increment of adult weight gain was 1.05 (95% CI, ) and 1.06 (95% CI, ) for CRC in North America and other areas, respectively. The pooled RR of the highest versus lowest categories and a 5-kg increment of adult weight gain was 1.65 (95% CI, ) and 1.10 (95% CI, ) in men for CRC, 1.07 (95% CI, ) and 1.02 (95% CI, ) in women for CRC, 1.60 (95% CI, ) and 1.10 (95% CI, ) in men for CC, 1.15 (95% CI, ) and 1.03 (95% CI, ) in women for CC, 1.44 (95% CI, ) and 1.08 (95% CI, ) in men for RC and 1.11 (95% CI, ) and 1.00 (95% CI, ) in women for RC, respectively. Discussion To our knowledge, this study is the first meta-analysis to evaluate the effect of adult weight gain on the risk of CRC. BMI and WC are the most commonly used measures to define obesity, which have been demonstrated to be risk factors for CRC regardless of tumor site or sex. 2,9,10 Whereas BMI and WC reflect both lean body mass and adipose mass, weight gain in adult life is generally a gain of body fat and thus potentially represents an age-related metabolic change that may be important in the development of CRC. 11 The findings from our meta-analysis indicate that adult weight gain is positively associated with the risk of CRC, CC and RC. However, the positive association is stronger among men than among women. Overall, the risk increases by 5% for

4 Table 1. Characteristic of the studies with regard to adult weight gain and risk of colorectal cancer Study Study design Source of study population Han et al., 2014, Cohort Atherosclerosis USA 12 Risk in Communities cohort Sample size/age (mean) Number of events Male% Definition of weight change 14,497/45 64 (NA) Age 25 years to baseline age Aleksandrova et al., 2013, Europe 14 Cohort European Prospective Investigation into Cancer and Nutrition 20,1696/NA (50) 2, Age 20 years to baseline age Bassett et al., 2010, Australia 15 Cohort Melbourne Collaborative Cohort Study 39,626/40 69 (55.7) Age 18 years to baseline age Type of cancer Weight change category (kg) RR (95% CI) CRC 23% to 3%, 10% 1 (reference) (F), 1.25 ( ) (F), 1 (reference) (M), 1.65 ( ) (M) CRC, CC, RC 22 to 2, 2 to 5, 5 to 10, 10 to 15, 15 to20, >20 1 (reference) (CRC), 1.24 ( ) (CRC), 1.12 ( ) (CRC), 1.13 ( ) (CRC), 1.31 ( ) (CRC), 1.37 ( ) (CRC) CC 23 to 3, 3 to 10, 10 to 20, 20 1 (reference) (F), 0.88 ( ) (F), 0.81 ( ) (F), 0.96 ( ) (F), 1 (reference) (M), 0.92 ( ) (M), 1.05 ( ) (M), 1.47 ( ) (M) Study quality 1 Adjustment 9 Age, race, center, education, height, smoking status at age 25, cigarette smoking status, alcohol consumption, physical activity at baseline and BMI at age 25 8 Age, center, weight at age 20, smoking status, education, alcohol intake, physical activity, red and processed meat, fish and shellfish, fruits and vegetables and fiber intake 8 Country of birth, weight at age 18, height at study entry, education, processed and fresh meat consumption, fruit and vegetable consumption, fat intake, daily energy intake, smoking status, alcohol consumption and sex

5 Table 1. Characteristic of the studies with regard to adult weight gain and risk of colorectal cancer (Continued) Study Study design Source of study population Oxentenko et al., Cohort Iowa Women s 2010, USA 20 Health Study Sample size/age (mean) 36,941/55 69 (62.1) Number of events Male% Definition of weight change 1,464 0 Age 18 years to baseline age Martinez et al., Cohort Nurses Health 1997, USA 19 Study 89,448/30 55 (NA) Age 18 years to baseline age Campbell et al., 2010, International 17 Case control Colon Cancer Family Registry 4,478/26 83 (56) 1, Age 20 to 1 year before CRC diagnosis Type of cancer Weight change category (kg) RR (95% CI) CRC 5, 5 to 12, 12 to 20, 20 1 (reference), 1.02 ( ), 1.16 ( ), 0.99 ( ) CC 25 to 5, 20 1 (reference), 1.08 ( ) CRC 0 to 5, 6 to 10, 11 to 20, 21 1 (reference), 1.09 ( ), 1.22 ( ), 1.48 ( ) Study quality 1 Adjustment 7 Age at baseline, baseline BMI, age at menopause, exogenous estrogen use, oral contraceptive use, smoking status, cigarette packyears, physical activity level, self-reported diabetes mellitus, intake of total energy, total fat, red meat, fruits and vegetables, calcium, folate, vitamin E and alcohol 7 Age, cigarette smoking, family history of colorectal cancer, leisure-time physical activity, postmenopausal hormone use, aspirin use, intake of red meat, alcohol consumption and BMI at age 18 7 Age, endoscopy screening, cigarette smoking, postmenopausal hormone use and BMI at age 20 years

6 Table 1. Characteristic of the studies with regard to adult weight gain and risk of colorectal cancer (Continued) Study Study design Source of study population Campbell et al., Case control Ontario Cancer 2007, Canada 16 Registry Sample size/age (mean) Number of events Male% Definition of weight change 5,364/20 74 (61.1) 2, Age 20 to 1 year before CRC diagnosis Type of cancer Weight change category (kg) RR (95% CI) CRC, CC, RC 1 to 5, 6 to 10, 11 to 20, 21 1 (reference) (F) (CRC), 1.05 ( ) (F) (CRC), 1.07 ( ) (F) (CRC), 1.13 ( ) (F) (CRC), 1 (reference) (M) (CRC), 1.06 ( ) (M) (CRC), 1.18 ( ) (M) (CRC), 1.64 ( ) (M) (CRC) Study quality 1 Adjustment 7 Age, education, red meat intake, recreational physical activity, province of residence, CRC screening endoscopy, history of high cholesterol/triglycerides, menopausal status and use of postmenopausal hormones Le Marchand et al., Case control Hawaii Tumor 1997, USA 18 Registry 2,384/NA (65) 1, Age 25 to 5 years prior to CRC diagnosis CRC Highest quartile vs. lowest quartile 1 (reference) (F), 0.80 (0.50 to 1.30) (F), 1 (reference) (M), 1.60 (1.00 to 2.40) (M) 7 Age, family history of colorectal cancer, alcoholic drinks/week, pack-years, egg, dietary fiber, calcium intakes, calories, lifetime recreational physical activity and BMI 5 years ago 1 Study quality was judged based on the Newcastle Ottawa Scale (range, 1 9 stars). Abbreviations: USA, United States of America; BMI, body mass index; CRC, colorectal cancer; CC, colon cancer; RC, rectal cancer; NA, not available; RR, risk ratio; CI, confidence interval; F, female; M, male.

7 2886 Adult weight gain and colorectal cancer Figure 2. Forest plots of relative risks of colorectal cancer for the highest versus lowest categories of adult weight gain. every 5-kg increase in adult weight for CRC, 6% for CC and 6% for RC. In sensitivity analyses and subgroup analyses, the results of cohort studies were consistent with the overall results after excluding case control studies, which means that there was no heterogeneity from study design. The three case control studies included in the meta-analysis were well conducted with explicit case definition, community controls and adequate covariates adjusted in the statistical analysis. Also, both in North America and other areas, there was a positive association between adult weight gain and risk of CRC, and the effect of adult weight gain on risk of CRC in most regions and ethnicities was consistent. Several mechanisms have been proposed to explain the association between adult weight gain and risk of CRC, including the insulin and insulin-like growth factor (IGF) axis, adiponectin and leptin, inflammation and steroid hormone. First, obesity was found to be related to insulin resistance and Type-2 diabetes mellitus,33 which reduces IGFbinding proteins levels and thereby increases free circulating IGF-I concentrations.34 Epidemiologic studies have indicated that high insulin concentrations35 and IGF-I concentrations36 C 2014 UICC Int. J. Cancer: 136, (2015) V

8 Chen et al Figure 3. Dose response relationship between adult weight gain and colorectal cancer risk. are positively associated with risk of CRC. Second, obesity is positively associated with serum leptin but inversely associated with adiponectin concentrations, 37 which has been shown to be associated with CRC in epidemiologic studies. 38,39 Third, excess secretion of cytokines from adipose tissues could induce a chronic proinflammatory response and thus promote CRC. 40 The sex difference in associations between adult weight gain and risk of CRC in our study might be explained by the influence of sex hormones. Obesity reduces testosterone concentrations in men 41 but increases testosterone concentrations in women. 42 Several studies have shown that high testosterone concentrations improve insulin sensitivity and low testosterone concentrations increase insulin resistance. 43,44 If insulin resistance is positively associated with risk of CRC, sex difference in associations between adult weight gain and CRC risk might be due to an obesity-induced reduction in testosterone concentrations. Also, adult weight gain that is not associated with risk of CRC might be explained by the influence of estrogens. Most of the female populations included in our meta-analysis are postmenopausal women except for one study conducted by Martinez et al., 19 which found a positive association between adult weight gain and risk of CRC in premenopausal women. Among postmenopausal women, the primary source of endogenous estrogen is through androgen conversion from adipocytes. 45 Exogenous estrogens (such as oral contraceptives and postmenopausal hormone therapy) have been associated with a decreased risk of CRC. 46,47 Therefore, in postmenopausal women, the protective effects of estrogen derived from adipose tissue may counter balance the otherwise risk-enhancing effects of obesity. 16 Our meta-analysis has several limitations. First, all of the included studies used long-term recall of body weight at early adulthood, which may introduce recall bias to the association between adult weight gain and risk of CRC. However, the accuracy of self-reports of past body weight has been generally supported in epidemiologic studies. 48,49 Second, definitions of adult weight gain in the included studies differed in life periods considered, which made meta-analysis somewhat difficult. Nevertheless, after study selection according to inclusion and exclusion criteria, all of the included studies used recalled weight between 18 and 25 years of age as the initial weight and used baseline weight in cohort studies and weight before diagnosis of CRC in case control studies as the ending weight. Finally, because thiswasameta-analysisofobservationalstudies,wecould not resolve problems with confounding factors that might be inherent in the included primary studies. Most of the included studies adjusted only for age and weight at early adolescence, but other important confounding factors were not adjusted in all included studies, such as diet factors, physical activity, social class, endoscopy screening, genetic factors and so on. In conclusion, there is evidence that adult weight gain is associated with an increased risk of CRC in men but not in women. Recognition of this risk is of great importance in both clinical practice and as a public health issue. Avoidance of weight gain needs to be emphasized throughout adulthood, and cancer prevention should be targeted according to sex. Future research should explore whether weight gain during specific life stages is a stronger predictor of risk of CRC than weight gain throughout adulthood and whether weight control by interventions in the adult population reduces the risk of CRC. Acknowledgments Q.C., J.W., J.Y. and J.H. discussed and developed the study question for this article. Q.C. and J.W. conducted the literature searches and assessed the eligibility of the studies for inclusion. Q.C. and J.Y. extracted data and

9 2888 Adult weight gain and colorectal cancer Table 2. Meta analyses of association between adulthood absolute weight change and risk of colorectal cancer Highest versus lowest Dose response analysis Colorectal cancer RR No. 1 (95% CI) I 2 (p-value) Egger test (p-value) No. 1 Linear model, per 5 kg Linearity test (p-value) 5 kg 2 10 kg 2 15 kg 2 20 kg 2 All studies ( ) 46.7% (0.04) ( ) ( ) 1.08 ( ) 1.14 ( ) 1.21 ( ) Exclusion of case control studies ( ) 14.3% (0.32) ( ) ( ) 1.04 ( ) 1.07 ( ) 1.12 ( ) Geographic area North America ( ) 53.9% (0.03) ( ) ( ) 1.07 ( ) 1.13 ( ) 1.18 ( ) Other areas ( ) 14.2% (0.32) ( ) ( ) 1.07 ( ) 1.14 ( ) 1.21 ( ) Sex Men ( ) 0.0% (0.42) ( ) ( ) 1.20 ( ) 1.32 ( ) 1.46 ( ) Women ( ) 0.0% (0.66) ( ) ( ) 1.04 ( ) 1.06 ( ) 1.08 ( ) Colon cancer All studies ( ) 41.4% (0.13) ( ) ( ) 1.04 ( ) 1.10 ( ) 1.19 ( ) Sex Men ( ) 0.0% (0.55) ( ) ( ) 1.09 ( ) 1.18 ( ) 1.33 ( ) Women ( ) 0.0% (0.54) ( ) ( ) 1.01 ( ) 1.04 ( ) 1.09 ( ) Rectal cancer All studies ( ) 17.3% (0.30) ( ) ( ) 1.05 ( ) 1.09 ( ) 1.15 ( ) Sex Men ( ) 0.0% (0.85) ( ) ( ) 1.13 ( ) 1.22 ( ) 1.32 ( ) Women ( ) 4.5% (0.31) ( ) ( ) 1.01 ( ) 1.01 ( ) 1.02 ( ) 1 If a single study reported results on different populations but did not report the overall results, the result for each was calculated as a different study 2 Adjusted RR of colorectal cancer when comparing 5 20 kg of adult weight gain with no weight gain, which was calculated from spline model Abbreviations: CI, confidence interval; RR, relative risk.

10 Chen et al conducted all analyses. All authors were involved in interpretation of the data and discussed the results. Q.C. wrote the first draft of this article, and it was reviewed by Z.J. and J.H. All authors agreed on the final draft of this article. J.H. is the guarantor. Reference 1. Ferlay J, Shin HR, Bray F, et al. Estimates of worldwide burden of cancer in 2008: GLOBO- CAN Int J Cancer 2010;127: Moghaddam AA, Woodward M, Huxley R. Obesity and risk of colorectal cancer: a meta-analysis of 31 studies with 70,000 events. Cancer Epidemiol Biomarkers Prev 2007;16: Sung JJ, Lau JY, Goh KL, et al. Increasing incidence of colorectal cancer in Asia: implications for screening. Lancet Oncol 2005;6: Cheng J, Chen Y, Wang X, et al. Meta-analysis of prospective cohort studies of cigarette smoking and the incidence of colon and rectal cancers. Eur J Cancer Prev Schmid D, Leitzmann MF. Association between physical activity and mortality among breast cancer and colorectal cancer survivors: a systematic review and meta-analysis. Ann Oncol 2014;25: Theodoratou E, Farrington SM, Tenesa A, et al. Associations between dietary and lifestyle risk factors and colorectal cancer in the Scottish population. Eur J Cancer Prev 2014;23: Bianchini F, Kaaks R, Vainio H. Overweight, obesity, and cancer risk. Lancet Oncol 2002;3: Ng M, Fleming T, Robinson M, et al. Global, regional, and national prevalence of overweight and obesity in children and adults during : a systematic analysis for the Global Burden of Disease Study Lancet 2014;384: Ma Y, Yang Y, Wang F, et al. Obesity and risk of colorectal cancer: a systematic review of prospective studies. PLOS One 2013;8:e Ning Y, Wang L, Giovannucci EL. A quantitative analysis of body mass index and colorectal cancer: findings from 56 observational studies. Obes Rev 2010;11: Ballard-Barbash R, Schatzkin A, Taylor PR, et al. Association of change in body mass with breast cancer. Cancer Res 1990;50: Han X, Stevens J, Truesdale KP, et al. Body mass index at early adulthood, subsequent weight change and cancer incidence and mortality. Int J Cancer 2014;135: Renehan AG, Flood A, Adams KF, et al. Body mass index at different adult ages, weight change, and colorectal cancer risk in the National Institutes of Health-AARP Cohort. Am J Epidemiol 2012;176: Aleksandrova K, Pischon T, Buijsse B, et al. Adult weight change and risk of colorectal cancer in the European Prospective Investigation into Cancer and Nutrition. Eur J Cancer 2013;49: Bassett JK, Severi G, English DR, et al. Body size, weight change, and risk of colon cancer. Cancer Epidemiol Biomarkers Prev 2010;19: Campbell PT, Cotterchio M, Dicks E, et al. Excess body weight and colorectal cancer risk in Canada: associations in subgroups of clinically defined familial risk of cancer. Cancer Epidemiol Biomarkers Prev 2007;16: Campbell PT, Jacobs ET, Ulrich CM, et al. Casecontrol study of overweight, obesity, and colorectal cancer risk, overall and by tumor microsatellite instability status. J Natl Cancer Inst 2010; 102: Le Marchand L, Wilkens LR, Kolonel LN, et al. Associations of sedentary lifestyle, obesity, smoking, alcohol use, and diabetes with the risk of colorectal cancer. Cancer Res 1997;57: Martinez ME, Giovannucci E, Spiegelman D, et al. Leisure-time physical activity, body size, and colon cancer in women. Nurses Health Study Research Group. J Natl Cancer Inst 1997; 89: Oxentenko AS, Bardia A, Vierkant RA, et al. Body size and incident colorectal cancer: a prospective study of older women. Cancer Prev Res (Phila) 2010;3: Wells G, Shea B, O Connell D, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. Ottawa, ON: Ottawa Hospital Research Institute, Available at: programs/clinical_epidemiology/oxford.asp. 22. Greenland S. Quantitative methods in the review of epidemiologic literature. Epidemiol Rev 1987;9: Orsini N, Bellocco R, Greenland S. Generalized least squares for trend estimation of summarized dose-response data. Stata J 2006;6: Greenland S, Longnecker MP. Methods for trend estimation from summarized dose-response data, with applications to meta-analysis. Am J Epidemiol 1992;135: DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7: Larsson SC, Orsini N, Wolk A. Dietary magnesium intake and risk of stroke: a meta-analysis of prospective studies. Am J Clin Nutr 2012;95: Larsson SC, Orsini N. Coffee consumption and risk of stroke: a dose-response meta-analysis of prospective studies. Am J Epidemiol 2011;174: Aune D, Greenwood DC, Chan DS, et al. Body mass index, abdominal fatness and pancreatic cancer risk: a systematic review and non-linear dose-response meta-analysis of prospective studies. Ann Oncol 2012;23: Jin Z, Xiang C, Cai Q, et al. Alcohol consumption as a preventive factor for developing rheumatoid arthritis: a dose-response meta-analysis of prospective studies. Ann Rheum Dis 2014;73: Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. BMJ 2003; 327: Egger M, Davey Smith G, Schneider M, et al. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315: Duval S, Tweedie R. Practical estimates of the effect of publication bias in meta-analysis. Australas Epidemiol 1998;5: Shoelson SE, Herrero L, Naaz A. Obesity, inflammation, and insulin resistance. Gastroenterology 2007;132: Simpson HL, Umpleby AM, Russell-Jones DL. Insulin-like growth factor-i and diabetes. A review. Growth Horm IGF Res 1998;8: Schoen RE, Tangen CM, Kuller LH, et al. Increased blood glucose and insulin, body size, and incident colorectal cancer. J Natl Cancer Inst 1999;91: Kaaks R, Toniolo P, Akhmedkhanov A, et al. Serum C-peptide, insulin-like growth factor (IGF)-I, IGF-binding proteins, and colorectal cancer risk in women. J Natl Cancer Inst 2000;92: Laughlin GA, Barrett-Connor E, May S. Sex-specific association of the androgen to oestrogen ratio with adipocytokine levels in older adults: the Rancho Bernardo Study. Clin Endocrinol 2006;65: Stattin P, Palmqvist R, Soderberg S, et al. Plasma leptin and colorectal cancer risk: a prospective study in Northern Sweden. Oncol Rep 2003;10: Wei EK, Giovannucci E, Fuchs CS, et al. Low plasma adiponectin levels and risk of colorectal cancer in men: a prospective study. J Natl Cancer Inst 2005;97: Riondino S, Roselli M, Palmirotta R, et al. Obesity and colorectal cancer: role of adipokines in tumor initiation and progression. World J Gastroenterol 2014;20: Muller M, den Tonkelaar I, Thijssen JH, et al. Endogenous sex hormones in men aged years. Eur J Endocrinol 2003;149: Oh JY, Barrett-Connor E, Wedick NM, et al. Endogenous sex hormones and the development of type 2 diabetes in older men and women: the Rancho Bernardo study. Diabetes Care 2002;25: Kapoor D, Goodwin E, Channer KS, et al. Testosterone replacement therapy improves insulin resistance, glycaemic control, visceral adiposity and hypercholesterolaemia in hypogonadal men with type 2 diabetes. Eur J Endocrinol 2006;154: Malkin CJ, Jones TH, Channer KS. The effect of testosterone on insulin sensitivity in men with heart failure. Eur J Heart Failure 2007;9: Simpson ER, Bulun SE, Nichols JE, et al. Estrogen biosynthesis in adipose tissue: regulation by paracrine and autocrine mechanisms. J Endocrinol 1996;150(Suppl):S51 S La Vecchia C, Gallus S, Fernandez E. Hormone replacement therapy and colorectal cancer: an update. J Br Menopause Soc 2005;11: Chlebowski RT, Wactawski-Wende J, Ritenbaugh C, et al. Estrogen plus progestin and colorectal cancer in postmenopausal women. N Engl J Med 2004;350: Kovalchik S. Validity of adult lifetime selfreported body weight. Public Health Nutr 2009; 12: Perry GS, Byers TE, Mokdad AH, et al. The validity of self-reports of past body weights by U.S. adults. 1995;6:61 6.