Prostate cancer is the second most common cancer among men globally (Ferlay et al., 2013) and the most common cancer among men in the USA today (American Cancer Society, 2014). Although incidence rates have increased during the previous decades, partly because of the implementation of prostate specific antigen (PSA) testing, they have stabilized on a high level over the past 10 years (Center et al., 2012). During the same time period, the prevalence of overweight and obesity has also increased worldwide (World Health Organization WHO, 2014).
The global prevalence of overweight and obesity among men has been estimated at 61% with even higher estimates in the USA, where 72% of adult men were overweight or obese in 2014 (World Health Organization WHO, 2014). Both overweight and obesity, often measured as BMI (kg/m2), and prostate cancer affect substantial proportions of the male population today. Body weight is related to metabolic and hormonal pathways, that may have an influence on tumorigenesis, and fat mass that plays an important role in the metabolism of sex hormones (Hsing et al., 2007). Prostate cancer is hormone dependent and likely linked to anthropometric factors that may affect, for example, pathways of insulin and insulin-like growth factors (IGFs), sex hormones, and adipokine signaling (Roberts et al., 2010). The potential impact of overweight and obesity on prostate cancer is therefore important to clarify.
Results from two previous meta-analyses have shown weak (MacInnis and English, 2006) and nonstatistically significant (Renehan et al., 2008) direct associations between BMI and overall risk of prostate cancer. The inconsistent and weak associations seen previously could possibly be explained by a nonlinear biological association between BMI and overall prostate cancer risk (Bhaskaran et al., 2014). However, heterogeneity of the disease may be another explanation. A meta-analysis by Discacciati et al. (2012) showed an inverse relationship between BMI and localized disease and a positive association between BMI and advanced prostate cancer. A more recent study also showed a decreased risk of low-grade prostate cancer and an increased risk of high-grade prostate cancer among obese men (Vidal et al., 2014). Although the association between BMI and prostate cancer risk appears complex, overweight and obesity prevention may provide a unique opportunity for primary prevention should a causal association be present.
The aim of the proposed study is to further investigate the association between BMI and risk of overall, aggressive and nonaggressive prostate cancer by examining the association between baseline BMI and subsequent prostate cancer risk among men in the Carotene and Retinol Efficacy Trial (CARET).
Participants and methods
CARET is a multicenter randomized, double-blind placebo-controlled chemoprevention trial and has previously been described in detail (Thornquist et al., 1993). Briefly, the trial tested whether daily supplementation with 30 mg β-carotene and 25 000 IU retinyl palmitate would reduce the risk of lung cancer among 18 314 heavy smokers, former smokers, and asbestos-exposed workers.
Beginning in 1985, participants were recruited from six centers across the USA. Eligible for participation were both men and women aged 50–69 years who were current or former smokers (within the previous 6 years) with a history of at least 20 pack-years of cigarette smoking [n=14 254 (55.9%) were male], and men aged 45–69 years (n=4060) who were current or former smokers (within the previous 15 years) with evidence of extensive occupational exposure to asbestos. CARET participants were recruited from health insurance rolls, managed care organizations, labor unions, workmen’s compensation programs, and by occupational physicians. Recruitment ended in 1994. Participants attended yearly clinic visits and completed follow-up telephone calls every 4 months throughout the intervention period that was stopped 21 months early in January 1996 when interim analysis found evidence that the supplements increased the risk of lung cancer and total mortality in this high-risk population (Omenn et al., 1996). Although the CARET intervention ended in 1996, active follow-up continued until 2005, including the collection of endpoint data for 94% of all randomized participants. This report is, however, restricted to male participants only.
The Institutional Review Board of the Fred Hutchinson Cancer Research Center and each of the five other participating institutions approved all procedures for the study. Participants provided written informed consent at recruitment.
All male CARET participants with the exclusion of those with a previous prostate cancer diagnosis before baseline (n=25) were included in the present study. Additional exclusions were made on the basis of the exposure and men with missing information on BMI (n=67), or a BMI less than 18 kg/m2 (n=45) or above 60 kg/m2 (n=2) were excluded as outliers. In total 11 886 men were included in final analysis. Additional self-reported information of physical activity was available from 6594 (55.5%) of participants.
Measured weight and height from each participant’s first CARET clinic visit was used to calculate BMI (kg/m2). Study participants were divided into categories of BMI 18–24.9, 25–29.9, 30–34.9, and at least 35 kg/m2 (Expert Panel on the Identification, Evaluation, and Treatment of Overweight in Adults, 1998). Information regarding age, sex, race/ethnicity, and general health history were also collected at the first clinic visit. Detailed data on smoking were collected at each CARET contact. This information consisted of current smoking status and smoking history, including age at smoking initiation, total years smoked, and average number of cigarettes smoked per day.
Dietary intake over the previous year was assessed at baseline and every 2 years using a self-administered food frequency questionnaire. For this study, the baseline assessment was used when available; if the baseline food frequency questionnaire was missing, the earliest available assessment was selected. Any reported energy intake less than 800 kcal/day or more than 5000 kcal/day were considered unreliable and therefore not included in analysis. Family history of prostate cancer was self-reported and defined as having at least one first degree relative with prostate cancer. Participants also self-reported level of education.
Physical activity was assessed in 1996–1997 using a self-administered physical activity questionnaire previously described by Alfano et al. (2004). In brief, participants were asked to report their average daily hours within five activity categories; (i) sleeping, (ii) sitting, (iii) light, (iv) moderate, and (v) vigorous physical activity. Time spent on weekdays and weekend days were reported separately and participants estimated their activity to the closest 0.25 h. Total weekly hours within each activity category were summarized by multiplying reported weekday hours by 5 and adding reported weekend hours multiplied by 2. In the present study, weekly hours reported spent in moderate and vigorous activities were summarized into a variable of moderate-to-vigorous physical activity.
Outcome assessment: prostate cancer cases
At each CARET annual visit as well as during the quarterly follow-up telephone calls, participants were asked to report whether they had been diagnosed with any new cancers. All endpoints, including prostate cancer, were verified by the CARET Endpoints committee. Prostate cancer was a secondary end-point of the trial. As part of this study, an augmented review of medical records and Surveillance Epidemiology and End Results cancer registry files was made to obtain data on Gleason score and stage of disease at diagnosis. Medical records, including pathology reports, surgical reports, hospital records, and scan and radiography reports were obtained. Gleason score information was abstracted from the pathology reports of the diagnosing pathologist. Approximately 92% of the prostate cancer cases were adjudicated and confirmed. Aggressive prostate cancer was defined in two ways: primarily as Gleason score of at least 7 and/or stage III/IV (distant) at diagnosis as in previous studies on prostate cancer in CARET (Neuhouser et al., 2007) and additionally by excluding men with a Gleason score of 3+4 from the previous definition. This was done as large differences on prostate cancer mortality has been seen for men with Gleason score 7 depending on whether the tumor was diagnosed as a 3+4 or 4+3 Gleason score (Stark et al., 2009). Men with a Gleason score of 4+3 had a more than doubled rate of mortality compared with men with a score of 3+4. Nonaggressive disease was defined as Gleason score of less than 7 and stage I/II at diagnosis. In total, 196 men did not have enough information, that is, had missing data for Gleason score or stage, to classify disease. Widespread PSA screening was not in use when CARET began and information on serum PSA were therefore not available.
Distributions and means of characteristics were studied across categories of BMI. Associations were tested by using one-way analysis of variance for continuous variables and χ2-test for categorical variables. The Kaplan–Meier method was used to analyze unadjusted incidence rates for overall prostate cancer risk and risk of nonaggressive and aggressive disease, respectively.
Cox proportional hazards regression models were used to estimate hazard ratios (HRs) with 95% confidence intervals (CIs) for prostate cancer incidence between different BMI categories. Time from inclusion to CARET to cancer diagnosis, death, or censoring at end of follow-up, whichever came first, was used as the underlying time scale. BMI 18–24.9 kg/m2 was used as the reference category. The proportional hazards assumption was tested using Schoenfeld’s residuals.
All Cox models were adjusted for age at enrollment as a continuous variable. A priori adjustments for additional specific CARET confounders were made for intervention arm, study center, race/ethnicity, and smoking at baseline (pack-years, continuous variable). The association between additional covariates (family history of prostate cancer, total energy intake, and education level) and the outcome was tested for by including covariates, one by one, in Cox proportional hazards models. Additional adjustments in multivariable models were thereafter made for family history of prostate cancer and total energy intake (kcal/day, continuous). Within the subgroup of study participants with data on physical activity (n=6594), Cox models were additionally adjusted for moderate-to-vigorous physical activity (h/week, continuous). Men diagnosed with prostate cancer during follow-up before the physical activity assessment (n=173) were excluded from these models. Further, in additional sensitivity analysis, all models were run including only men in the subgroup with information on physical activity. To test for linear trends, the median value of each BMI category was used as a numerical variable in the Cox models. To examine whether preclinical symptoms of cancer might have affected BMI at baseline, leading to biased results, we excluded the first 2 years of follow-up from all analysis in sensitivity analysis.
P values were considered significant if less than 0.05. All statistical analyses were performed using STATA 13.0 (Stata Corporation, College Station, Texas, USA).
Characteristics of study participants are presented in Table 1. Overall, 23.0% of men had a BMI 18–24.9 kg/m2, 47.9% a BMI of 25–29.9 kg/m2, and 29.1% of men had a BMI of at least 30 kg/m2 at baseline. Significant differences between men in the different BMI categories were observed for age at baseline, smoking, total energy intake, moderate-to-vigorous physical activity, race/ethnicity, education level, and study center. During the follow-up, 883 (7.4%) men were diagnosed with prostate cancer. Of these, 353 (40.0%) were classified as aggressive disease when all Gleason 7 were included, 202 (22.9%) as aggressive when Gleason 3+4 was excluded from the definition, and 334 (37.8%) were classified as nonaggressive disease. The mean age at diagnosis was 67.5 (±5.9) years. The mean follow-up time for all men was 11.4 (±3.9) years. The average follow-up times to diagnosis for all prostate cancer cases, men diagnosed with aggressive, aggressive excluding Gleason 3+4, and nonaggressive disease were 7.2 (±3.9), 6.8 (±3.7), 6.5 (±3.9), and 6.7 (±3.9) years, respectively. The follow-up times differed significantly between all men and men diagnosed with prostate cancer during the follow-up (P<0.0001), but not between men diagnosed with aggressive and nonaggressive disease (P=0.78 and 0.51).
Incidence rates for overall, aggressive, aggressive excluding Gleason 3+4, and nonaggressive prostate cancer for all men in the cohort were 6.5, 2.5, 1.5, and 2.5 per 1000 person-years, respectively. Kaplan–Meier curves of unadjusted incidence rates for overall prostate cancer risk and risk of nonaggressive and aggressive prostate cancer, respectively, are shown in Fig. 1.
Results from age-adjusted and multivariable adjusted Cox proportional hazards models are presented in Table 2. No statistically significant differences were seen in incidence rates between the BMI categories for overall, aggressive, or nonaggressive prostate cancer. In analysis of aggressive prostate cancer when Gleason 3+4 was not included in the definition, a significantly increased incidence rate of disease was seen among men with BMI of at least 35 kg/m2 compared with the reference group with a BMI of 18–24.9 kg/m2 in both age-adjusted (HR: 1.78, 95% CI: 1.05–3.01), and multivariable adjusted models (HR: 1.77, 95% CI: 1.04–3.00, and HR: 1.80, 95% CI: 1.04–3.11).
Additional adjustment for physical activity restricted to only men with complete information from the physical activity questionnaire changed estimates slightly, but no statistically significant results were seen (data not shown). However, point estimates for aggressive disease not including Gleason 3+4 in the definition still indicate an increased incidence rate, although not statistically significant, among men in the highest BMI category (HR: 2.17, 95% CI: 0.84–5.57, Ptrend=0.17). Further, in sensitivity analysis where all models were restricted to only the subgroup with available information on physical activity, BMI estimates were more or less unchanged (data not shown). Results were also similar in sensitivity analysis where the first 2 years of follow-up had been excluded (data not shown). In addition, a linear trend was seen across BMI categories for aggressive disease excluding Gleason 3+4 (Ptrend=0.04). Analysis did not indicate any linear trends across BMI categories for overall, aggressive disease when including all Gleason 7, or nonaggressive prostate cancer.
Our results showed an increased incidence rate of aggressive prostate cancer among men in CARET with a baseline BMI of at least 35 kg/m2 compared with men with a BMI of 18–24.9 kg/m2 when Gleason score 3+4 was not included in the definition of aggressive disease. We did not demonstrate any associations between baseline BMI and risk of overall, aggressive (including all Gleason 7), or nonaggressive prostate cancer. The results are similar to those of previous meta-analyses that have shown only modest or nonsignificant associations between BMI and the risk of overall and less-aggressive prostate cancer (MacInnis and English, 2006; Renehan et al., 2008). Although Renehan et al. (2008) did not distinguish between localized and advanced disease in their systematic overview and meta-analysis, MacInnis and English (2006) did. Our results when not including men with Gleason 3+4 in the definition of aggressive disease are in line with the significantly increased risk of advanced disease shown with increasing BMI (relative risk: 1.12, 95% CI: 1.01–1.23 per 5 kg/m2 increase in BMI) presented by MacInnis and English (2006). In addition, an increased risk of advanced prostate cancer with increasing BMI was also seen in a meta-analysis by Discacciati et al. (2012) (relative risk: 1.08, 95% CI: 1.04–1.12). In contrast to our findings, Discacciati et al. (2012) also showed a significantly decreased risk of localized disease with increasing BMI, which is not supported in the present study. Our results further support the conclusion made in a report from the World Cancer Research Fund International stating that results from published studies constituted strong evidence for an increased risk of advanced prostate cancer among overweight and obese men (World Cancer Research Fund International/American Institute for cancer Research, 2014).
In this study, aggressive disease was defined in two different ways; first, as Gleason score of at least 7 and/or stage III/IV (distant) at diagnosis, and second, by additionally excluding men with a Gleason score of 3+4 from the first definition. The first definition has been used in previous studies on prostate cancer in CARET, whereas the second definition was included to take into account previous research indicating that cancer mortality may differ between men with a Gleason 3+4=7 compared with men with a Gleason 4+3=7. In Gleason 4+3 tumors, the prevalent pattern is that of poorly formed, fused, and cribriform glands, whereas the most prevalent pattern in 3+4 tumors comprise well formed, individual glands. Studies have shown that a Gleason score 4+3 demonstrates both worse pathological stage and biochemical recurrence rates than Gleason 3+4 (Gordetsky and Epstein, 2016). Gleason 4+3 cancers have also been associated with a more than doubled to three-fold risk of lethal prostate cancer compared with Gleason 3+4 cancers (Stark et al., 2009). Our results indicate that there might be a difference between Gleason 7 tumors depending on whether pattern 3 or 4 is more prevalent and that a distinction of the different Gleason 7 categories may be important.
Adiposity affects both metabolic and hormonal pathways that may have an influence on tumorigenesis and the differentiation of tumor cells. Obesity can lead to altered levels of adipokines, with increased levels of leptin and decreased levels of adiponectin, which have been associated with tumor development (Roberts et al., 2010). Increased adiposity has also been associated with hyperinsulinemia that, through reduction of IGF-binding proteins, leads to increased serum levels of IGF-1 that may promote tumor development (Renehan et al., 2006). An explanation to differences in risk estimates between different BMI categories and different prostate cancer subtypes may also be an effect of BMI on PSA levels. Higher BMI has been consistently associated with decreased levels of serum PSA (Baillargeon et al., 2005; Banez et al., 2007), potentially leading to later detection among men in today’s era of frequent PSA testing. Nevertheless, a decreased risk of low-grade prostate cancer and an increased risk of high-grade prostate cancer have also been shown among obese men independent of PSA level (Vidal et al., 2014).
There are several strengths and weaknesses of the present study that need to be acknowledged. The strengths include standardized data collection and excellent follow-up of participants during and after the trial with 92% of prostate cancer cases being adjudicated and confirmed. Further, the measured weight and height to compute BMI is also a strength. Nevertheless, BMI is nonspecific and does not differentiate between fat free mass and fat mass and individuals with a high muscle mass and low fat mass may therefore be incorrectly classified as overweight and obese (Prentice and Jebb, 2001; Okorodudu et al., 2010). In addition, BMI does not consider distribution and therefore cannot differentiate between more metabolically active abdominal fat and other fat (Schwenzer et al., 2010). Fat mass has been shown to play an important role in hormone metabolism that might influence tumorgenesis (Hsing et al., 2007). However, the limitations of BMI as a proxy for body composition are inherent to all studies using BMI if other measures of body composition are not available. Another limitation to this present study is the fact that CARET began before widespread use of PSA screening was implemented and information on serum PSA was not available to distinguish between screen-detected and clinically-detected cases. The low sample size in analysis of low-grade and high-grade prostate cancer, respectively, because of missing data on Gleason score and stage, which creates uncertainty in estimates, is also a limitation. Nevertheless, information on tumor stage is likely missing at random.
Further, although detailed assessment of several lifestyle factors enabled controlling for confounders, residual or unmeasured confounding cannot be ruled out. The collection of information on physical activity on average about 5 years after baseline is another limitation. In addition, physical activity (PA) was self-reported and there is no assurance that even if a PA measure was available on the entire cohort, that it would be an assessment free of bias and measurement error. Nevertheless, although PA levels are generally stable in this age group (Smith et al., 2015) and the collected data therefore are a reasonable approximation of usual PA also at baseline, men responding to the PA questionnaire are conditioned on being alive at the time of response. Therefore, selection bias cannot be ruled out.
As the CARET cohort is composed of current or former heavy smokers, the generalizability of results to the general population may be limited. In addition, the limited follow-up time restricted analysis to prostate cancer incidence and did not allow for analysis with mortality as an end-point. It has been suggested that BMI may play a greater role in the progression of prostate cancer tumors rather than in tumor development (Mistry et al., 2007) and a high BMI has been associated with an increased risk of both progression and mortality (Cao and Ma, 2011).
Our results indicate an increased risk of aggressive prostate cancer among men with a high BMI when Gleason 3+4 were not included in the definition of aggressive disease, but do not support an association between BMI and risk of overall, aggressive disease including all Gleason 7, or nonaggressive prostate cancer risk within a cohort of current and former heavy smokers. Further studies should investigate differences between different Gleason 7 patterns.
This work was supported by the National Cancer Institute, National Institute of Health U01-CA63673, UM1-CA167462 and R01-CA96789. Additional funding was also received from the Robert Lundberg Memorial Foundation and the Gålö Foundation.
Conflicts of interest
There are no conflicts of interests.
Alfano CM, Klesges RC, Murray DM, Bowen DJ, McTiernan A, Vander Weg MW, et al (2004). Physical activity in relation to all-site and lung cancer incidence and mortality in current and former smokers. Cancer Epidemiol Biomarkers Prev 13:2233–2241.
American Cancer Society (2014). Cancer facts & figures 2014. Atlanta: American Cancer Society.
Baillargeon J, Pollock BH, Kristal AR, Bradshaw P, Hernandez J, Basler J, et al (2005). The association of body mass index
and prostate-specific antigen in a population-based study. Cancer 103:1092–1095.
Banez LL, Hamilton RJ, Partin AW, Vollmer RT, Sun L, Rodriguez C, et al (2007). Obesity-related plasma hemodilution and PSA concentration among men with prostate cancer. JAMA 298:2275–2280.
Bhaskaran K, Douglas I, Forbes H, dos-Santos-Silvia I, Leon DA, Smeeth L (2014). Body-mass index and risk of 22 specific cancers: a population-based cohort study of 5.24 million UK adults. Lancet 384:755–765.
Cao Y, Ma J (2011). Body-mass index, prostate cancer-specific mortality and biochemical recurrence: a systematic review and meta-analysis. Cancer Prev Res 4:486–501.
Center MM, Jemal A, Lortet-Tieulent J, Ward E, Ferlay J, Brawley O, et al (2012). International variation in prostate cancer incidence and mortality rates. Eur Urol 61:1079–1092.
Discacciati A, Orsini N, Wolk A (2012). Body mass index
and incidence of localized and advanced prostate cancer: a dose–response meta-analysis of prospective studies. Ann Oncol 23:1665–1671.
Expert Panel on the Identification, Evaluation, and Treatment of Overweight in Adults (1998). Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults: executive summary. Am J Clin Nutr 68:899–917.
Ferlay J, Ervik M, Dikshit R, Eser S, Mathers C, Rebelo M, et al (2013) GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 11 [Internet]. International Agency for Research on Cancer. Avaliable at: http://globocan.iarc.fr
. [Accessed 15 May 2017].
Gordetsky J, Epstein J (2016). Grading of prostatic adenocarcinoma: current state and prognostic implications. Diagn Pathol 11:25.
Hsing AW, Sakoda LC, Chua S Jr. (2007). Obesity, metabolic syndrome, and prostate cancer. Am J Clin Nutr 86:843–857.
MacInnis RJ, English DR (2006). Body size and composition and prostate cancer risk
: systematic review and meta-regression analysis. Cancer Causes Control 17:989–1003.
Mistry T, Digby JE, Desai KM, Randeva HS (2007). Obesity and prostate cancer: a role for adipokines. Eur Urol 52:46–53.
Neuhouser ML, Barnett MJ, Kristal AR, Ambrosone CB, King I, Thornquist M, et al (2007). (n-6) PUFA increase and dairy foods decrease prostate cancer risk
in heavy smokers. J Nutr 137:1821–1827.
Okorodudu DO, Jumean MF, Montori VM, Romero-Corral A, Somers VK, Erwin PJ, et al (2010). Diagnostic performance of body mass index
to identify obesity as defined by body adiposity: a systematic review and meta-analysis. Int J Obes 34:791–799.
Omenn GS, Goodman GE, Thornquist MD, Balmes J, Cullen MR, Glass A, et al (1996). Risk factors for lung cancer and for intervention effects in CARET, the Beta-Carotene and Retinol Efficacy Trial. J Natl Cancer Inst 88:1550–1559.
Prentice AM, Jebb SA (2001). Beyond body mass index
. Obes Rev 2:141–147.
Renehan AG, Frystyk J, Flyvberg A (2006). Obesity and cancer risk: the role of the insulin-IGF axis. Trends Endocr Metab 17:328–336.
Renehan AG, Tyson M, Egger M, Heller RF, Zwahlen M (2008). Body-mass index and incidence of prostate cancer: a systematic review and meta-analysis of prospective observational studies. Lancet 371:569–578.
Roberts DL, Dive C, Renehan AG (2010). Biological mechanisms linking obesity and prostate cancer risk
: new perspectives. Annu Rev Med 61:301–316.
Schwenzer NF, Machann J, Schraml C, Springer F, Ludescher B, Stefan N, et al (2010). Quantitative analysis of adipose tissue in single transverse slices for estimation of volumes if relevant fat tissue compartments: a study in a large cohort of subjects at risk for type 2 diabetes by MRI with comparison to anthropometric data. Invest Radiol 45:788–794.
Smith L, Gardner B, Fisher A, Hamer M (2015). Patterns and correlates of physical activity behaviour over 10 years in older adults: prospective analyses from the English Longitudinal Study of Ageing. BMJ Open 5:e007423.
Stark JR, Perner S, Stampfer MJ, Sinnott JA, Finn S, Eisenstein AS, et al (2009). Gleason score and lethal prostate cancer: does 3+4=4+3? J Clin Oncol 27:3459–3464.
Thornquist MD, Omenn GS, Goodman GE, Grizzle JE, Rosenstock L, Barnhart S, et al (1993). Statistical design and monitoring of the Carotene and Retinol Efficacy Trial (CARET). Cont Clin Trials 14:308–324.
Vidal AC, Howard LE, Moreira DM, Castro-Santamaria R, Andriole GL Jr., Freedland SJ (2014). Obesity increases the risk for high-grade prostate cancer: results from the REDUCE study. Cancer Epidemiol Biomarkers Prev 23:2936–2942.
World Health Organization (WHO) (2014). Global Status Report on nuncommunicable diseases. Geneva: WHO.