Abstract
Background
The association of fitness with cancer risk is not clear.
Methods
We used Cox proportional hazards models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for risk of lung, colorectal, endometrial, breast, and prostate cancer in a subset of UK Biobank participants who completed a submaximal fitness test in 2009-12 (N = 72,572). We also investigated relationships using two-sample Mendelian randomisation (MR), odds ratios (ORs) were estimated using the inverse-variance weighted method.
Results
After a median of 11 years of follow-up, 4290 cancers of interest were diagnosed. A 3.5 ml O2⋅min−1⋅kg−1 total-body mass increase in fitness (equivalent to 1 metabolic equivalent of task (MET), approximately 0.5 standard deviation (SD)) was associated with lower risks of endometrial (HR = 0.81, 95% CI: 0.73–0.89), colorectal (0.94, 0.90–0.99), and breast cancer (0.96, 0.92–0.99). In MR analyses, a 0.5 SD increase in genetically predicted O2⋅min−1⋅kg−1 fat-free mass was associated with a lower risk of breast cancer (OR = 0.92, 95% CI: 0.86–0.98). After adjusting for adiposity, both the observational and genetic associations were attenuated.
Discussion
Higher fitness levels may reduce risks of endometrial, colorectal, and breast cancer, though relationships with adiposity are complex and may mediate these relationships. Increasing fitness, including via changes in body composition, may be an effective strategy for cancer prevention.
Similar content being viewed by others
Introduction
Until recently epidemiological studies have largely focused on the role of physical activity behaviours with cancer risk [1]. Cardiorespiratory fitness (referred to here as ‘fitness’) is distinct from physical activity as it describes the capacity of the circulatory and respiratory systems to supply oxygen to skeletal muscle during prolonged physical activity [2, 3]. Fitness is generally objectively measured and has a stronger genetic component than habitual physical activity [2,3,4].
Higher fitness is associated with good cardiometabolic health, including lower visceral adipose tissue, inflammation and insulin sensitivity, and may, therefore, reduce the risk of cancer [5,6,7,8]. Previous studies report that people with higher fitness have lower risks of all-cause mortality, cancer mortality and cardiovascular disease [5, 9,10,11], but the relationship between fitness and incident cancers are less clear. Some studies have reported inverse associations between fitness and lung and colorectal cancers [12,13,14,15,16,17], while for prostate cancer associations have been reported to be null or positive [13,14,15, 17,18,19,20]. Only one prior study has investigated associations between fitness and female-specific incident cancers, and did not find evidence of a relationship [14].
A limitation of observational epidemiological studies includes the possibility of residual confounding and reverse causation. Mendelian randomisation (MR) uses germline genetic variants as proxies of biological traits to generate instrumental variables and estimate their associations with disease risk. Because germline genetic variants are fixed and randomly allocated at conception, this technique may be less likely to be affected by biases and confounding factors (such as preclinical disease and smoking history). This is the first study to use MR to investigate fitness and cancer risk.
We aimed to assess the associations of measured fitness and risk of common cancers (lung, colon, rectal, endometrial, female breast, and prostate cancer) using observational methods in the UK Biobank. In secondary analyses, we used a two-sample MR framework, using genetically predicted fitness, as an instrumental variable derived from UK Biobank [21] and genetic case control data from consortia for those same sites, plus pancreatic cancer and renal cell carcinoma for which observational analyses in the UK Biobank are underpowered. By integrating evidence from both observational epidemiology and MR approaches, we aim to strengthen the basis for causal inference [22].
Methods
UK Biobank study population
The UK Biobank study is a population-based prospective cohort study of 502,625 adults aged 40 to 69 years. A description of the study protocol is available online [23]. Participants were registered with the UK National Health Service and lived within 40 km of a UK Biobank assessment centre in England, Wales, and Scotland. Baseline data were collected between 2006 and 2010. A repeat-measures substudy was conducted between 2012 and 2013.
UK Biobank cardiorespiratory fitness assessment
An individualised submaximal cycle ergometer test was implemented in 2009 and offered to 75,087 participants during baseline data collection, 17,109 participants during the repeat assessment study, and 2877 participants at both timepoints; 97,950 tests were offered in total. For those participants who were offered a test at both timepoints, the earliest fitness test completed by the participant was used to maximise follow-up duration. Participant baseline data were collected on the same day as their exercise test. The test was individualised to each participant’s exercise capacity and risk level for engaging in exercise. Participants with lower exercise capacity or higher risk for exercise-related complications were offered a test with lower work rates, while those with higher exercise capacity or lower risk were offered a test with higher work rates. A description of the exercise test individualisation process and maximal oxygen consumption (VO2 max; ml O2⋅min−1⋅kg−1) estimation process is provided in Supplementary Methods; the test protocol is available online [24]. VO2 max was estimated in two ways: scaled by total-body mass (VO2maxtbm [3.5 ml O2⋅min−1⋅kg−1 total-body mass=1 MET]) and scaled by fat-free mass (VO2maxffm) [25, 26]. VO2maxffm represents the ability of skeletal muscle to use oxygen during maximal exercise, whereas VO2maxtbm is more representative of aerobic performance capacity [27].
Genetic instrument for cardiorespiratory fitness
Full details of the fitness genome-wide association study (GWAS) are available elsewhere [21]. In brief, single nucleotide polymorphisms (SNPs) associated with fitness were identified from a GWAS based on UK Biobank participants of European ancestry who participated in the fitness test (N included = 69,416). Fitness was estimated using the same framework method described above, scaled by fat-free mass and using resting heart rate data from the full cohort, excluding those taking beta-blockers (N included = 452,941) (P < 5 × 10−8 significance threshold).
The Radial plot method was used to select eligible resting heart-rate associated genetic variants for fitness by removing heterogeneous outliers for the genetic variants, of which 149 were also nominally significant in the fitness GWAS (p < 0.05) [28]. The genetic instrument for fitness included 14 fitness and 149 fitness and resting heart rate variants with prioritisation given to the variants identified in the fitness GWAS. In total, 160 independent (r² > 0.01) genetic variants were included in our instrument for fitness [21].
Cancer ascertainment
Observational analysis
Cancer registration data were provided via record linkage to national cancer and death registries, until the following censoring dates: 31 July 2019 in England and Wales and 31 October 2015 in Scotland. Cancers occurring after the registry censoring dates were identified using Hospital Episode Statistics (HES), until the following censoring dates: 30 September 2021 in England, 31 July 2021 in Scotland and 28 February 2018 in Wales (see Supplementary Methods for cancer site definitions).
Of the 84,792 fitness tests analysed after preliminary exclusions (i.e., participant withdrawal of data, ‘high risk’ for exercise; see Supplementary Fig. 1), we retained a preliminary analytic sample of 79,347 participants after additionally excluding 3209 participants for missing data, 1017 due to test data quality, 1219 with missing weight, fat-free mass, or heart rate, and 44 for whom fitness estimation could not be applied. We then excluded 5180 participants with prevalent cancer at baseline and 1551 participants diagnosed with cancer within two years of follow-up. The final analytic sample was 72,572 participants. Health and sociodemographic characteristics were described across age-adjusted and sex-specific fitness tertiles [29].
Genetic cancer data
Risk estimates may be biased when instrumental variables and outcomes are identified from the same sample [30]. We, therefore, used independent GWAS data from international consortia. This includes breast (including estrogen receptor (ER)+ and ER− subtypes) [31, 32], prostate (including aggressive disease) [33], endometrial [34], ovarian [35], lung (including for never smokers) [36], and colorectal cancer (including colon, rectal, male colorectal and female colorectal, distal colon and proximal colon) [37, 38]. We also included pancreatic cancer and renal cell carcinoma [39,40,41,42]. Included sites and subtypes were chosen based on data availability. Further information for the genetic case control studies is available in Supplementary Table 1.
Statistical analysis
Observational analysis
Cox regression models with age as the underlying timescale were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) per 3.5 ml⋅O2⋅min−1⋅kg−1 total-body mass and 5.0 ml⋅O2⋅min−1⋅kg−1 fat-free mass for risk of cancer diagnosis. Models were adjusted for possible confounding factors and for female reproductive cancers (breast, endometrial and ovarian cancers) we additionally adjusted for reproductive factors (see Supplementary Methods). Multivariate imputation by chained equations was used to impute missing covariate values.
Adiposity may partially mediate and confound the relationship between fitness and cancer risk (Supplementary Fig. 2). Therefore, we evaluated the role of adiposity in fitness-to-cancer associations both with and without adjustment for either BMI (for models with VO2max scaled by total-body mass) or fat mass (for models with VO2max scaled by fat-free mass).
We have shown previously that repeat assessments of the UK Biobank fitness test will elicit moderately stable fitness estimates (regression dilution ratio = 0.79, standard error = 0.01) [43]. This source of measurement error will influence the strength of observed health associations. Therefore, in a sensitivity analysis, we also provide regression dilution calibrated estimates of fitness-to-cancer associations using established statistical techniques [44]. The shape of dose-response relationships between fitness and risk of cancer diagnosis was investigated using cubic spline regression models. Each model used two knots placed at the 33rd and 67th percentile of the fitness distribution. Reference values were set to the mean fitness value for each specific analysis (see Supplementary Figs. 3 and 4).
Sensitivity analysis
Subgroup analyses for colorectal cancer were examined by sex, and associations for fitness and lung cancer were re-examined after restricting the analysis to never-smokers only. Subgroups were chosen a priori on the basis of data availability and previous evidence for heterogeneity in the associations [1]. We also included a minimally adjusted model to investigate the influence of mediators and/or confounders.
Mendelian randomisation
The MR estimation for fitness and cancer was conducted using the inverse-variance weighted (IVW) method [45]. We additionally calculated the I2GX statistic to assess measurement error in SNP-exposure associations [33], the F-statistic to examine the strength of the genetic instrument [46], Cochran’s Q statistic for heterogeneity between the MR estimates for each SNP [47], and PhenoScanner was used to assess pleiotropy of the genetic instruments [48]. As sensitivity analyses, we used the MR residual sum and outlier (MR-PRESSO) to investigate the role of SNP outliers [49]. To assess pleiotropy, we used the weighted median and contamination mixture methods [50].
To explore relationships between body fat and fitness, we conducted a bi-directional MR of genetically predicted fitness on fat mass and vice versa using our genetic instrument for fitness and an instrument for total fat mass based on a GWAS of UK Biobank participants (N = 330,762 participants of European ancestry), derived from bioelectrical impedance measurements at study baseline [51]. We also conducted multivariable MR (MVMR) analyses to assess the effect of fitness on cancer risk, after accounting for genetically predicted fat mass and height [45].
Statistical software
Observational analyses were performed using Stata version 16.1 (Stata Corporation, College Station, TX, USA). MR analyses were performed using the TwoSampleMR and MendelianRandomisation R packages [52, 53] and figures were plotted in R version 3.6.3. All tests of significance were two-sided, and P < 0.05 were considered statistically significant. Results are presented in accordance with the STROBE checklist [54].
Results
Observational analysis
After a median of 11 years of follow-up, 1586 prostate cancers, 1093 breast cancers, 811 colorectal cancers, 480 lung cancers, 184 endometrial cancers, and 136 ovarian cancers were diagnosed. Participant characteristics by age-adjusted and sex-specific fitness tertiles are provided in Table 1 for fitness scaled by total-body mass and Supplementary Table 2 for fitness scaled by fat-free mass. Fitness was higher in men compared to women, and those in the middle and higher fitness tertiles had better measures of adiposity, socioeconomic status, and cardiometabolic health than those in the lower fitness tertile.
Observational analysis results are summarised in Fig. 1. In analyses without BMI adjustment, each 3.5 ml O2⋅min−1⋅kg−1 total-body mass increase (equivalent to 1 metabolic equivalent of task [MET]) in fitness was associated with a 19% reduction in endometrial cancer, 6% reduction in colorectal cancer, and 4% reduction in breast cancer. After BMI adjustment, associations were attenuated but remained directionally consistent. Where associations were detected, relationships generally appeared to be linear but with uncertainty for some cancers at the tails of the fitness distribution (Supplementary Figs. 3 and 4). When fitness was expressed per kg fat-free mass, associations with cancers were not significant. Results adjusted for regression dilution are shown in Supplementary Fig. 5.
HRs and 95% CIs estimated using Cox regression models adjusted for age, sex, self-reported racial/ethnic group, Townsend index of deprivation, education, employment status, smoking status, alcohol consumption, red and processed meat consumption, fish consumption, fruit and vegetable consumption, salt consumption, diabetes status, hypertension, medication use (beta blockers, calcium channel blockers, ACE inhibitors, diuretics, bronchodilators, lipid-lowering agents, iron deficiency agents, non-steroidal anti-inflammatory drugs, metformin). Female reproductive cancers (breast, endometrial, and ovarian) were additionally adjusted for age at menarche, age at menopause, parity, hormone replacement therapy usage, and oral contraceptives. Associations with and without adjustment for either continuous BMI (for models with VO2max scaled by total-body mass) or fat mass (for models with VO2max scaled by fat-free mass). ACE Angiotensin-converting enzyme, BMI body mass index, CI confidence interval, HR hazard ratio.
There was evidence of heterogeneity in the associations of fitness and colorectal cancers by sex; the relationship was inverse for men and null for women (Fig. 2 and Supplementary Fig. 4). Minimally adjusted models are available from Supplementary Table 3.
HRs and 95% CIs estimated using Cox regression models adjusted for age, sex, self-reported racial/ethnic group, Townsend index of deprivation, education, employment status, smoking status, alcohol consumption, red and processed meat consumption, fish consumption, fruit and vegetable consumption, salt consumption, diabetes status, hypertension, medication use (beta blockers, calcium channel blockers, ACE inhibitors, diuretics, bronchodilators, lipid-lowering agents, iron deficiency agents, non-steroidal anti-inflammatory drugs, metformin). Associations with and without adjustment for either continuous BMI (for models with VO2max scaled by total-body mass) or fat mass (for models with VO2max scaled by fat-free mass). ACE Angiotensin-converting enzyme, BMI body mass index, CI confidence interval, HR hazard ratio.
Mendelian randomisation analyses
Higher levels of genetically predicted fitness were associated with a lower risk of breast cancer (OR per 5.0 ml O2⋅min−1⋅kg−1 fat-free mass = 0.92, 95% CI: 0.86–0.98; P = 0.02), including ER+ (0.91, 0.84–0.99; P = 0.02) and ER- (0.88, 0.80–0.97; P = 0.01) subtypes, but was not significantly associated with any other cancer site (Fig. 3). There was also no evidence of an association with colorectal cancer after stratification by sex and site (Supplementary Tables 4 and 5). There was significant heterogeneity in the MR estimates for the SNPs for each cancer site (Cochran’s Q P < 0.05), except for associations with lung cancer for never smokers (P = 0.13), aggressive prostate cancer (P = 0.17) and renal cancer (P = 0.09).
Associations were estimated using the inverse variance weighted method. CI confidence interval, ER estrogen receptor, OR odds ratio.
In MR sensitivity analyses, the relationships between fitness and breast cancer were directionally consistent in comparison with the primary MR analysis (Supplementary Table 6). There was evidence of an inverse association between fitness and lung cancer using the weighted median method (0.85, 0.74–0.98; P = 0.02) and a positive association with pancreatic cancer using the contamination mixture method (1.09, 1.03–1.14; P = 0.03) (Supplementary Table 6). Radial plots also did not indicate any strong influence of outliers on the MR results (Supplementary Fig. 6). The likelihood of bias due to weak instruments was low (F-statistic > 10 for all SNPs). There was evidence of moderate levels of measurement error (I2GX = 0.52–0.65), indicating reduced reliability of Egger results, therefore we do not include Egger estimates [55]. Using PhenoScanner, 742 traits were linked to SNPs for fitness (P < 5 × 10−8), particularly pulse rate (Supplementary Fig. 7).
The bi-directional MR analysis indicated that genetically instrumented fat mass had a strong inverse association with fitness (OR per 0.5 SD increase = 0.61, 0.52–0.71; P < 0.001), but a weaker inverse relationship of fitness with fat mass (OR per 5 ml O2⋅min−1⋅kg−1 fat-free mass = 0.96, 0.92–1.01; P = 0.08). In MVMR analyses, associations with breast cancer were attenuated after adjustment for fat mass and height. While associations with lung cancer became statistically significant (0.90, 0.84–0.96; P = 0.002), although remained null for never smokers (Table 2).
Discussion
This study used both observational and MR methods to examine the relationship between cardiorespiratory fitness and incident cancer risk, providing the first evidence that higher fitness levels may reduce risks of breast cancer. In observational analysis only, we report additional inverse associations between VO2max scaled to total body mass and risks of colorectal and endometrial cancer. However, associations with all three cancer sites were attenuated after accounting for adiposity. Observational associations between cancer and VO2max scaled to fat-free mass were not statistically significant.
Previous observational analyses have reported inverse associations between fitness and colorectal and lung cancer. We did not observe an association with lung cancer and the inverse association between VO2max scaled to total body mass and colorectal cancer was attenuated after accounting for BMI [12,13,14,15]. Our results may differ from these previous studies due to differences in population sampling, fitness assessment, and fitness estimation approaches. For example, cycle ergometer-based fitness estimates may differ from treadmill-based estimates due to differences in load bearing and motion artefact [15, 18, 20]. The UK Biobank fitness test was also relatively light intensity, which enabled more participants to be assessed. Thus, our analysis likely characterises a wider variety of lower-fitness individuals than previous studies which used more strenuous tests. Previous estimates using UK Biobank data had shorter duration of follow-up (median 5 years) and used fewer exercise test data, which will reduce the precision of risk estimates.
Previous MR studies based on up to five SNPs have reported inverse associations between genetically predicted physical activity levels and risks of breast, colorectal and aggressive prostate cancer [56, 57]. However, current estimates suggest that GWAS significant polymorphisms explain a very limited proportion of phenotypic physical activity (e.g., 0.06% for overall physical activity) [58]. The small number of SNPs increase the influence of possibly invalid variants within the instrument, and the instrument has a bidirectional association with BMI [58]. Fitness is a trait that reflects both input from genetics and physical activity behaviours. The genetic instrument for fitness used in the present study likely encompasses both past and current levels as well as the capacity to participate in physical activity [2, 3]. This instrument explains 1.2% of the variation in observed fitness levels, increasing the reliability of risk estimates. Future work examining the relative importance of the different constituents of genetic fitness may help to clarify whether the null relationships that we report for fitness on colorectal and aggressive prostate cancer risk are indicative of the greater relative importance of physical activity behaviours or are partially reflective of the methodological limitations discussed above.
The role of adiposity in fitness is complex and not fully understood. Higher adiposity is associated with impaired physical performance, relating reduced muscle oxygen uptake, lower cardiac efficiency, neuromuscular dysfunction, and increased cancer risk [59,60,61,62,63]. Higher levels of physical activity are important for weight maintenance and increasing fitness, and higher fitness may reduce some of the harmful cardiometabolic effects of obesity [64]. Differences between the associations of fitness and cancer by scaling are likely driven by the different components of fitness, as VO2maxtbm has a strong inverse correlation with body size and adiposity [27]. However, the complex interplay of adiposity, fitness and cancer might mean that accounting for adiposity for models of cardiorespiratory fitness could lead to an over-adjustment of risk estimates, but these relationships are difficult to disentangle. Relationships between fitness and all-cause, cancer, and cardiovascular mortality outcomes has stronger evidence for independence of associations with adiposity [10, 64,65,66]. Future work with longer durations of follow-up will improve power to investigate whether there are differential risk associations by BMI classification.
These analyses have several strengths. This study is the first to use genetically instrumented fitness to evaluate possible causal relationships between fitness and cancer risk. The UK Biobank is the largest sample currently available with measured cardiorespiratory fitness, maximising power to assess associations across a broad range of cancer sites, the majority of which have not been previously investigated. Our independently validated novel framework to estimate fitness harmonised the UK Biobank test protocols and calibrated these data to a maximal exercise test to estimate VO2max. This estimation framework also incorporated multiple heart rate measurements to reduce measurement error, with high temporal agreement (regression dilution ratio=0.79) over approximately a 2.8 year period for greater precision in risk estimates [43]. Further, the baseline assessment collected data across a wide range of lifestyle, medical and anthropometric factors, enabling thorough adjustment for possible confounders.
Our study has limitations. This analysis is not a randomised controlled trial and therefore we are not able to fully assess causality. In MR analysis we cannot exclude the possibility of genetic confounding or horizontal pleiotropy [67]. The genetic instrument for VO2maxtbm was not available for comparison with our observational analysis. The genetic instrument also included resting heart rate information; therefore, our results may be partially driven by genetic associations with resting heart rate. Given the strong a priori evidence and mechanistic plausibility of associations between fitness and cancer risk we have not included correction for multiple testing [18,19,20], however, we cannot exclude the possibility of chance findings. The UK Biobank participants are predominantly of White European ancestry and are healthier than the underlying sampling population, therefore risk estimates may not be generalisable to some other populations, including “high-risk” participants who did not undergo the fitness assessment. The fitness test was also submaximal, which may increase measurement error, and previous studies have noted larger magnitudes of associations with health outcomes using maximal fitness tests [11, 15].
In summary, we provide evidence that higher fitness levels may reduce risks of endometrial, colorectal, and breast cancer. The role of adiposity in mediating the relationship between fitness and cancer risk is not fully understood, and further research is needed to explore this complex relationship. Aiming to increase fitness, including via changes in body composition, may be an effective strategy to reduce risk of some cancer sites.
Data availability
The UK Biobank is an open-access resource and bona fide researchers can apply to use the UK Biobank dataset by registering and applying at http://ukbiobank.ac.uk/register-apply/. Further information is available from the corresponding author upon request. Fitness estimates for each participant have been returned to UK Biobank for download by registered researchers. For genetic cancer data, summary GWAS statistics are publicly available for breast (https://bcac.ccge.medschl.cam.ac.uk/), endometrial (https://www.ebi.ac.uk/gwas/studies/GCST006464), ovarian (https://ocac.ccge.medschl.cam.ac.uk) and prostate (overall only) (http://practical.icr.ac.uk/). Data for renal cell carcinoma and pancreatic cancer were accessed via dbGaP; Study Accession: phs000206.v3.p2 and phs000648.v1.p1; project reference 9314 (https://www.ncbi.nlm.nih.gov/gap/). Summary genetic data for aggressive prostate cancer (http://practical.icr.ac.uk/), lung cancer (https://ilcco.iarc.fr/) and colorectal cancer (https://research.fredhutch.org/peters/en/genetics-and-epidemiology-of-colorectal-cancer-consortium.html) are not currently publicly available but may be made available upon application, see respective websites for details.
Code availability
Analyses were performed using Stata version 16.1 (Stata Corporation, College Station, TX, USA) and R (version 4.3.1) the code is available upon reasonable request.
References
Moore SC, Lee IM, Weiderpass E, Campbell PT, Sampson JN, Kitahara CM, et al. Association of leisure-time physical activity with risk of 26 types of cancer in 1.44 Million adults. JAMA Intern Med. 2016;176:816–25.
Sarzynski MA, Ghosh S, Bouchard C. Genomic and transcriptomic predictors of response levels to endurance exercise training. J Physiol. 2017;595:2931–9.
Kim DS, Wheeler MT, Ashley EA. The genetics of human performance. Nat Rev Genet. 2022;23:40–54.
Bouchard C, An P, Rice T, Skinner JS, Wilmore JH, Gagnon J, et al. Familial aggregation of VO2max response to exercise training: results from the HERITAGE Family Study. J Appl Physiol. 1999;87:1003–8.
Al-Mallah MH, Sakr S, Al-Qunaibet A. Cardiorespiratory fitness and cardiovascular disease prevention: an update. Curr Atheroscler Rep. 2018;20:1.
Kaze AD, Agoons DD, Santhanam P, Erqou S, Ahima RS, Echouffo-Tcheugui JB. Correlates of cardiorespiratory fitness among overweight or obese individuals with type 2 diabetes. BMJ Open Diabetes Res Care. 2022;10:e002446.
LaMonte MJ, Eisenman PA, Adams TD, Shultz BB, Ainsworth BE, Yanowitz FG. Cardiorespiratory fitness and coronary heart disease risk factors: the LDS Hospital Fitness Institute cohort. Circulation. 2000;102:1623–8.
Arsenault BJ, Lachance D, Lemieux I, Alméras N, Tremblay A, Bouchard C, et al. Visceral adipose tissue accumulation, cardiorespiratory fitness, and features of the metabolic syndrome. Arch Intern Med. 2007;167:1518–25.
Han M, Qie R, Shi X, Yang Y, Lu J, Hu F, et al. Cardiorespiratory fitness and mortality from all causes, cardiovascular disease and cancer: dose–response meta-analysis of cohort studies. Br J Sports Med [Internet]. 2022. [cited 2022 May 13]; Available from: https://bjsm.bmj.com/content/early/2022/01/12/bjsports-2021-104876.
Schmid D, Leitzmann MF. Cardiorespiratory fitness as predictor of cancer mortality: a systematic review and meta-analysis. Ann Oncol. 2015;26:272–8.
Laukkanen JA, Isiozor NM, Kunutsor SK. Objectively assessed cardiorespiratory fitness and all-cause mortality risk: an updated meta-analysis of 37 cohort studies involving 2,258,029 participants. Mayo Clin Proc. 2022;97:1054–73.
Marshall CH, Al-Mallah MH, Dardari Z, Brawner CA, Lamerato LE, Keteyian SJ, et al. Cardiorespiratory fitness and incident lung and colorectal cancer in men and women: results from the Henry Ford Exercise Testing (FIT) cohort. Cancer. 2019;125:2594–601.
Pozuelo-Carrascosa DP, Alvarez-Bueno C, Cavero-Redondo I, Morais S, Lee IM, Martínez-Vizcaíno V. Cardiorespiratory fitness and site-specific risk of cancer in men: a systematic review and meta-analysis. Eur J Cancer. 2019;113:58–68.
Steell L, Ho FK, Sillars A, Petermann-Rocha F, Li H, Lyall DM, et al. Dose-response associations of cardiorespiratory fitness with all-cause mortality and incidence and mortality of cancer and cardiovascular and respiratory diseases: the UK Biobank cohort study. Br J Sports Med. 2019;53:1371–8.
Lakoski SG, Willis BL, Barlow CE, Leonard D, Gao A, Radford NB, et al. Midlife cardiorespiratory fitness, incident cancer, and survival after cancer in men: the cooper center longitudinal study. JAMA Oncol. 2015;1:231–7.
Hillreiner A, Baumeister SE, Sedlmeier AM, Finger JD, Schlitt HJ, Leitzmann MF. Association between cardiorespiratory fitness and colorectal cancer in the UK Biobank. Eur J Epidemiol. 2020;35:961–73.
Ekblom-Bak E, Bojsen-Møller E, Wallin P, Paulsson S, Lindwall M, Rundqvist H, et al. Association between cardiorespiratory fitness and cancer incidence and cancer-specific mortality of colon, lung, and prostate cancer among Swedish Men. JAMA Netw Open. 2023;6:e2321102.
Vainshelboim B, Müller J, Lima RM, Nead KT, Chester C, Chan K, et al. Cardiorespiratory fitness and cancer incidence in men. Ann Epidemiol. 2017;27:442–7.
Kunutsor SK, Voutilainen A, Laukkanen JA. Cardiorespiratory fitness is not associated with reduced risk of prostate cancer: a cohort study and review of the literature. Eur J Clin Invest. 2021;51:e13545.
Byun W, Sui X, Hébert JR, Church TS, Lee IM, Matthews CE, et al. Cardiorespiratory fitness and risk of prostate cancer: findings from the Aerobics Center Longitudinal Study. Cancer Epidemiol. 2011;35:59–65.
Cai L, Gonzales T, Wheeler E, Kerrison N, Day F, Langenberg C, et al. Causal associations between cardiorespiratory fitness and type 2 diabetes. Nat Commun. 2023;14:3904.
Lawlor DA, Tilling K, Davey Smith G. Triangulation in aetiological epidemiology. Int J Epidemiol. 2016;45:1866–86.
UK Biobank: Protocol for a large-scale prospective epidemiological resource [Internet]. 2007. Available from: https://www.ukbiobank.ac.uk/media/gnkeyh2q/study-rationale.pdf.
UK Biobank cardio assessment manual: Version 1.0 [Internet]. 2011. Available from: http://biobank.ctsu.ox.ac.uk/crystal/docs/Cardio.pdf.
Imboden MT, Kaminsky LA, Peterman JE, Hutzler HL, Whaley MH, Fleenor BS, et al. Cardiorespiratory fitness normalized to fat-free mass and mortality risk. Med Sci Sports Exerc. 2020;52:1532–7.
Osman AF, Mehra MR, Lavie CJ, Nunez E, Milani RV. The incremental prognostic importance of body fat adjusted peak oxygen consumption in chronic heart failure. J Am Coll Cardiol. 2000;36:2126–31.
Zhou N. Assessment of aerobic exercise capacity in obesity, which expression of oxygen uptake is the best? Sports Med Health Sci. 2021;3:138–47.
Bowden J, Spiller W, Del Greco MF, Sheehan N, Thompson J, Minelli C, et al. Improving the visualization, interpretation and analysis of two-sample summary data Mendelian randomization via the Radial plot and Radial regression. Int J Epidemiol. 2018;47:1264–78.
Kokkinos P, Myers J, Franklin B, Narayan P, Lavie CJ, Faselis C. Cardiorespiratory fitness and health outcomes: a call to standardize fitness categories. Mayo Clin Proc. 2018;93:333–6.
Haycock PC, Burgess S, Wade KH, Bowden J, Relton C, Davey Smith G. Best (but oft-forgotten) practices: the design, analysis, and interpretation of Mendelian randomization studies1. Am J Clin Nutr. 2016;103:965–78.
Michailidou K, Lindström S, Dennis J, Beesley J, Hui S, Kar S, et al. Association analysis identifies 65 new breast cancer risk loci. Nature. 2017;551:92–4.
Zhang H, Ahearn TU, Lecarpentier J, Barnes D, Beesley J, Qi G, et al. Genome-wide association study identifies 32 novel breast cancer susceptibility loci from overall and subtype-specific analyses. Nat Genet. 2020;52:572–81.
Schumacher FR, Al Olama AA, Berndt SI, Benlloch S, Ahmed M, Saunders EJ, et al. Association analyses of more than 140,000 men identify 63 new prostate cancer susceptibility loci. Nat Genet. 2018;50:928–36.
O’Mara TA, Glubb DM, Amant F, Annibali D, Ashton K, Attia J, et al. Identification of nine new susceptibility loci for endometrial cancer. Nat Commun. 2018;9:3166.
Phelan CM, Kuchenbaecker KB, Tyrer JP, Kar SP, Lawrenson K, Winham SJ, et al. Identification of 12 new susceptibility loci for different histotypes of epithelial ovarian cancer. Nat Genet. 2017;49:680–91.
McKay JD, Hung RJ, Han Y, Zong X, Carreras-Torres R, Christiani DC, et al. Large-scale association analysis identifies new lung cancer susceptibility loci and heterogeneity in genetic susceptibility across histological subtypes. Nat Genet. 2017;49:1126–32.
Huyghe JR, Bien SA, Harrison TA, Kang HM, Chen S, Schmit SL, et al. Discovery of common and rare genetic risk variants for colorectal cancer. Nat Genet. 2019;51:76–87.
Huyghe JR, Harrison TA, Bien SA, Hampel H, Figueiredo JC, Schmit SL, et al. Genetic architectures of proximal and distal colorectal cancer are partly distinct. Gut. 2021;70:1325–34.
Scelo G, Purdue MP, Brown KM, Johansson M, Wang Z, Eckel-Passow JE, et al. Genome-wide association study identifies multiple risk loci for renal cell carcinoma. Nat Commun. 2017;8:15724.
Childs EJ, Mocci E, Campa D, Bracci PM, Gallinger S, Goggins M, et al. Common variation at 2p13.3, 3q29, 7p13 and 17q25.1 associated with susceptibility to pancreatic cancer. Nat Genet. 2015;47:911–6.
Petersen GM, Amundadottir L, Fuchs CS, Kraft P, Stolzenberg-Solomon RZ, Jacobs KB, et al. A genome-wide association study identifies pancreatic cancer susceptibility loci on chromosomes 13q22.1, 1q32.1 and 5p15.33. Nat Genet. 2010;42:224–8.
Amundadottir L, Kraft P, Stolzenberg-Solomon RZ, Fuchs CS, Petersen GM, Arslan AA, et al. Genome-wide association study identifies variants in the ABO locus associated with susceptibility to pancreatic cancer. Nat Genet. 2009;41:986–90.
Gonzales TI, Westgate K, Strain T, Hollidge S, Jeon J, Christensen DL, et al. Cardiorespiratory fitness assessment using risk-stratified exercise testing and dose-response relationships with disease outcomes. Sci Rep. 2021;11:15315.
Keogh RH, White IR. A toolkit for measurement error correction, with a focus on nutritional epidemiology. Stat Med. 2014;33:2137–55.
Burgess S, Bowden J. Integrating summarized data from multiple genetic variants in Mendelian randomization: Bias and coverage properties of inverse-variance weighted methods [Internet]. arXiv; 2015 [cited 2022 May 13]. Available from: http://arxiv.org/abs/1512.04486.
Burgess S, Thompson SG, CHD CRP. Genetics collaboration. Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol. 2011;40:755–64.
Burgess S, Bowden J, Fall T, Ingelsson E, Thompson SG. Sensitivity analyses for robust causal inference from Mendelian randomization analyses with multiple genetic variants. Epidemiol Camb Mass. 2017;28:30–42.
Kamat MA, Blackshaw JA, Young R, Surendran P, Burgess S, Danesh J, et al. PhenoScanner V2: an expanded tool for searching human genotype-phenotype associations. Bioinforma Oxf Engl. 2019;35:4851–3.
Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet. 2018;50:693–8.
Bowden J, Davey, Smith G, Haycock PC, Burgess S. Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator. Genet Epidemiol. 2016;40:304–14.
Neale lab [Internet]. [cited 2022 Aug 2]. UK Biobank. Available from: http://www.nealelab.is/uk-biobank.
Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-Base platform supports systematic causal inference across the human phenome. eLife. 2018;7:e34408.
Yavorska OO, Burgess S. MendelianRandomization: an R package for performing Mendelian randomization analyses using summarized data. Int J Epidemiol. 2017;46:1734–9.
Mansournia MA, Collins GS, Nielsen RO, Nazemipour M, Jewell NP, Altman DG, et al. A checklist for statistical assessment of medical papers: the CHAMP statement. Br J Sports Med. 2021;55:1002–3.
Bowden J, Del Greco MF, Minelli C, Davey Smith G, Sheehan NA, Thompson JR. Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression: the role of the I2 statistic. Int J Epidemiol. 2016;45:1961–74.
Kazmi N, Haycock P, Tsilidis K, Lynch BM, Truong T. PRACTICAL consortium, CRUK, BPC3, CAPS, PEGASUS, et al. Appraising causal relationships of dietary, nutritional and physical-activity exposures with overall and aggressive prostate cancer: two-sample Mendelian-randomization study based on 79,148 prostate-cancer cases and 61,106 controls. Int J Epidemiol. 2020;49:587–96.
Papadimitriou N, Dimou N, Tsilidis KK, Banbury B, Martin RM, Lewis SJ, et al. Physical activity and risks of breast and colorectal cancer: a Mendelian randomisation analysis. Nat Commun. 2020;11:597.
Doherty A, Smith-Byrne K, Ferreira T, Holmes MV, Holmes C, Pulit SL, et al. GWAS identifies 14 loci for device-measured physical activity and sleep duration. Nat Commun. 2018;9:5257.
Wearing SC, Hennig EM, Byrne NM, Steele JR, Hills AP. The impact of childhood obesity on musculoskeletal form. Obes Rev Off J Int Assoc Study Obes. 2006;7:209–18.
Hung TH, Liao PA, Chang HH, Wang JH, Wu MC. Examining the relationship between cardiorespiratory fitness and body weight status: empirical evidence from a population-based survey of adults in Taiwan. ScientificWorldJournal. 2014;2014:463736.
Tanner CJ, Barakat HA, Dohm GL, Pories WJ, MacDonald KG, Cunningham PRG, et al. Muscle fiber type is associated with obesity and weight loss. Am J Physiol Endocrinol Metab. 2002;282:E1191–6.
Norman AC, Drinkard B, McDuffie JR, Ghorbani S, Yanoff LB, Yanovski JA. Influence of excess adiposity on exercise fitness and performance in overweight children and adolescents. Pediatrics. 2005;115:e690–6.
Larsson SC, Spyrou N, Mantzoros CS. Body fatness associations with cancer: evidence from recent epidemiological studies and future directions. Metabolism. 2022;137:155326.
Barry VW, Baruth M, Beets MW, Durstine JL, Liu J, Blair SN. Fitness vs. fatness on all-cause mortality: a meta-analysis. Prog Cardiovasc Dis. 2014;56:382–90.
Farrell SW, Cortese GM, LaMonte MJ, Blair SN. Cardiorespiratory fitness, different measures of adiposity, and cancer mortality in men. Obesity. 2007;15:3140–9.
Sui X, LaMonte MJ, Laditka JN, Hardin JW, Chase N, Hooker SP, et al. Cardiorespiratory fitness and adiposity as mortality predictors in older adults. JAMA J Am Med Assoc. 2007;298:2507–16.
Davies NM, Holmes MV, Smith GD. Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ. 2018;362:k601.
Acknowledgements
This research has been conducted using the UK Biobank Resource under application number #408 and #9905. The authors have no conflicts of interest to disclose. This work utilised the computational resources of the NIH HPC Biowulf cluster. (http://hpc.nih.gov). We thank all participants, researchers and support staff who made the study possible. We additionally would like to thank BCAC, ECAC, GECCO, ILCCO, OCAC, PanScan, PanC4 and the PRACTICAL consortia for their invaluable contributions to the genetic analysis. Genetic consortium specific acknowledgements and funding information are available in Appendix 1. The findings and conclusions in this article are those of the authors and do not necessarily represent the official position of the National Institutes of Health.
Funding
EW, SM, CM and PSM are supported by the Intramural Research Programme of the National Institutes of Health (NIH). NW, TS, FRD, KW, TG and SB are supported by UK Medical Research Council [grant numbers MC_UU_00006/1, MC_UU_00006/2 and MC_UU_00006/4]. NW and SB are supported by the NIHR Biomedical Research Centre in Cambridge (IS-BRC-1215-20014). The NIHR Cambridge Biomedical Research Centre (BRC) is a partnership between Cambridge University Hospitals NHS Foundation Trust and the University of Cambridge, funded by the National Institute for Health Research (NIHR). The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.
Author information
Authors and Affiliations
Consortia
Contributions
ELW, TIG, PFSM, NJW, CEM, SCM, SB conceptualised the study. TIG and ELW designed and conducted the analysis. TIG conducted the observational analysis and created framework to estimate fitness. ELW conducted the MR analysis. NJW, SB, TIG and FRD contributed to the fitness GWAS. DTB, SJC, MJ, TOK, LLM, VM, PAN, CCN, RKP, The PRACTICAL consortium, CRUK, BPC3, CAPS, PEGASUS, MPP, CMU, KSB, BVG contributed genetic cancer data. ELW, TIG, PFSM, NJW, CEM, SCM, SB, TS, FRD and KW were involved in drafting the manuscript. All authors contributed to the interpretation of results, and to the writing and revision of the manuscript. TIG had full access to all the observational data and ELW had full access to the genetic data. They take responsibility for the integrity of the data and the accuracy of the data analysis.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Ethics approval and consent to participate
UK Biobank received ethical approval from the National Health Service National Research Ethics Service (Ref 11/NW/0382). Participants studies gave informed consent. This study was performed in accordance with the Declaration of Helsinki.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Watts, E.L., Gonzales, T.I., Strain, T. et al. Observational and genetic associations between cardiorespiratory fitness and cancer: a UK Biobank and international consortia study. Br J Cancer 130, 114–124 (2024). https://doi.org/10.1038/s41416-023-02489-3
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41416-023-02489-3
