Associations of Individual and Combined Physical Activity and Body Mass Index Groups with Proinflammatory Biomarkers among Colorectal Cancer Patients

This article is featured in Highlights of This Issue, p. 2093

Energy balance components, including physical activity and body mass index (BMI), are established risk and predictive factors of colorectal cancer (1–4). The mechanistic underpinnings of the energy balance–colorectal cancer link are complex and have yet to be fully elucidated. A thorough understanding of the underlying molecular pathways may provide the opportunity to identify targets to intervene and improve colorectal cancer prevention, treatment, and survivorship.

Various biological mechanisms have been investigated to understand the energy balance–cancer link (e.g., immune function, oxidative stress, inflammation, angiogenesis, growth factors, and the gut microbiome; refs. 3–5). As one of the important hallmarks of cancer (6), systemic inflammation has been identified as a potential key player (3). Adipose tissue in obese individuals leads to systemic low-grade inflammation through the secretion of proinflammatory cytokines by hypertrophic adipocytes (3, 7–10). Visceral adipose tissue depots are considered the main source of obesity-induced proinflammatory processes (9). Physical activity, on the other hand, has been associated with reduced systemic inflammation (11–22). Exercise accompanied with weight loss leads to greater reductions in systemic inflammation; yet, exercise without weight loss also improves proinflammatory biomarker levels such as C-reactive protein (CRP; refs. 14, 18, 23). Previous studies have been limited to healthy individuals, smaller panels of biomarkers, or were conducted in restricted study populations (e.g., female, breast cancer), and results may not be directly generalizable to other types of cancers (11–22).

Another emerging question in the energy balance–cancer link is whether or not physical activity can counteract the adverse metabolic profiles commonly observed among obese individuals. “Metabolically healthy obese” patients are obese individuals who do not present the phenotype characterized by obesity, including systemic inflammation and insulin resistance (24–26). No standardized definition of a “metabolically healthy obese” phenotype exists (24, 25). Components of metabolic syndrome (blood glucose levels, HDL cholesterol levels, triglycerides, etc.) with and without inflammation are considered important factors differentiating metabolically healthy and unhealthy obese. Some data suggest that “metabolically healthy obese” individuals are generally younger and more active (24, 25). That raises the question of whether or not physical activity can offset the deleterious effects of obesity-induced inflammation. This question has never been studied in a cancer population, where inflammation is a key player of disease development and progression.

The objective of this study is to assess associations of individual and combined physical activity and BMI groups with proinflammatory biomarkers [CRP, serum amyloid A (SAA), interleukin 6 (IL6), interleukin 8 (IL8), and tumor necrosis factor alpha (TNFα)] in colorectal cancer patients.

The present study is conducted as part of the prospective, multicenter ColoCare Study (ClinicalTrials.gov NCT02328677), an international cohort of newly diagnosed stage I–IV colorectal cancer patients (ICD-10 C18–C20; ref. 27). The ColoCare Study design has previously been described (27, 28). Briefly, the ColoCare Study is a multicenter cohort to facilitate transdisciplinary research on colorectal cancer outcomes and prognosis. Patients who meet the following inclusion criteria are approached at the participating recruitment sites about 2 to 4 weeks before undergoing surgery: patients first diagnosed with colon or rectal cancer (stages I–IV), age >18 years, English (U.S. sites) or German (German site) speaking, and mentally/physically able to consent and participate. Participants were staged according to the American Joint Committee on Cancer (AJCC) system based on histopathologic findings. Overall, the recruitment rate was 70%. Baseline assessments include biospecimen collection, and patient-reported symptoms, health behaviors, and health-related quality of life, assessed by questionnaire. All analyses in this article are based on data collected from 579 patients enrolled between June 2007 and July 2019 at the ColoCare Study sites at the National Center for Tumor Diseases (NCT) and University of Heidelberg (HD), Heidelberg in Germany and the Huntsman Cancer Institute (HCI), Salt Lake City, Utah, and the Fred Hutchinson Cancer Center (FHCC), Seattle, Washington, with available baseline blood samples. The study was approved by the institutional review boards of the respective institutions and was conducted in accordance with the 1964 Helsinki Declaration. All patients provided written informed consent.

Non-fasting blood samples were collected from patients prior to surgery. Serum was extracted within four hours after blood draw and stored in aliquots at −80°C until analysis. 110 to 500 μL of each patient's serum enrolled at the ColoCare site in Germany and FHCC in the United States was shipped on dry ice to HCI, Salt Lake City, Utah, for laboratory analyses. Blood samples used in this study were stored on average for 3 to 5 years (HCI samples: <1–3 years, HD samples: 2–6 years, and FHCC: 8–14 years). Prior studies reported high stability of the analyzed biomarkers over long-term storage (3–6 years), limiting the possibility of measurement error (29–35). We further conducted a sensitivity analysis excluding samples that were stored for more than 6 years (FHCC samples) to see whether or not long-term storage influences the observed results. No differences in results were observed.

Serum-based assays for multiplexed CRP, SAA, IL6, IL8, and TNFα have previously been established on the Mesoscale Discovery Platform (MSD) in the Ulrich laboratory at HCI (27, 36–40). The biomarker panel selection was based on (i) our own research/preliminary data using circulating biomarkers as prognostic markers in colorectal cancer patients (36, 39, 41); (ii) the existing literature (11–18, 42–46); and (iii) a choice of the most clinically relevant biomarkers (41, 47–50).

Blinded patient samples plus three intraplate and interplate quality control samples were assayed for CRP and SAA using the V-PLEX vascular injury plate 2, and for IL6, IL8, and TNFα using the U-PLEX proinflammatory custom plate for samples from the HD cohort and V-PLEX proinflammatory custom plate for samples from the HCI and FHCC cohorts. Assays were conducted on the Sector 2400A (MSD).

Blinded serum samples were run at dilutions of 1:1,000 (HD samples) and 1:2,500 (HCI and FHCC) for the vascular injury panel and at dilutions of 1:2 for custom proinflammatory panels. The serum samples had no previous freeze–thaw cycles for the vascular injury panels and one freeze–thaw cycle for the proinflammatory panels. Data were analyzed with MSD Discovery Workbench 4.0 software (MSD). Drift correction was applied to account for batch effects across sites using the batch with the highest coefficients of variability (CV) as the reference batch. The overall inter- and intraplate CVs after drift correction were between 7.9–14.6% and 2.1–4.8%, respectively (27, 36–40).

Recreational physical activity within the past year before diagnosis was assessed using an adapted version of the International Physical Activity Questionnaire (IPAQ) questionnaire (51). Multiple choice questions were used instead of the text entry questions in the original questionnaire version. Patients were asked if they walked for exercise, did moderate or vigorous exercise for at least 10 minutes over the past year (“no,” “yes, less than once a week,” “yes, more than once a week”). If they responded “yes, more than once a week,” patients were asked about how many days per week (1–2, 3–4, 5–7), and how many minutes per day (10–29, 30–44, 45–59, 60+). In addition, patients were asked about their usual walking pace (casual, moderate, and fast). The IPAQ questionnaire captures data on usual recreational physical activity during the preceding year (51). The summation of duration (hours) and frequency (per week) of moderate to vigorous activity can be calculated in metabolic equivalent tasks hours per week (MET-hours/week) for each patient to determine the average amount of time per week that the patient spent in moderate to vigorous physical activity. The assignment of MET values follows the most recent Compendium of Physical Activities (52) and the questionnaire has previously been validated in a large international cohort (53). The assessment of moderate to vigorous physical activity using the IPAQ instrument has been validated as compared with objective measurements and other self-reported measurements (53).

BMI (kg/m²) at baseline (presurgery) was abstracted from patient medical charts. We conducted a blinded review of a subset of charts (10%) across sites to ensure quality of the data abstraction. For the purposes of this study, underweight patients (BMI ≤18.5 kg/m²) were excluded from the analyses due to the small sample size in this subgroup (n = 10) and to minimize confounding by patients with poor prognosis (54, 55).

Moderate activity was defined as 3.5 to 6 MET-hours and vigorous activities as ≥6 MET-hours (56). Thus, 8.75 MET-hours/week would be the threshold to meet the guidelines of at least 150 minutes (=2.5 hours) per week of moderate to vigorous activity as recommended for cancer survivors (57, 58). Accordingly, patients were categorized into either “inactive” (<8.75 MET-hours/week) or “active” (≥8.75 MET-hours/week) categories to test the effect of physical activity levels below and above the guidelines. For sensitivity analyses, patients were grouped into four physical activity groups to investigate a dose response: <75 minutes/week, 75–150 minutes/week, 150–300 minutes/week, and >300 minutes/week. Patients were categorized as being “normal weight” (BMI: ≥18.5 and <25 kg/m²), “overweight (BMI: ≥25 and <30 kg/m²), or “obese” (BMI: ≥25 kg/m²; ref. 59). To ensure sufficient sample sizes in each group, patients were categorized into “normal weight” and “overweight/obese” for analyses that tested combinations with physical activity groups. Combining physical activity and BMI information, patients were further categorized into: (i) “normal weight/active” (≥18.5 and <25 kg/m², ≥8.75 MET-hours/week); (ii) “normal weight/inactive” (≥18.5 and <25 kg/m², <8.75 MET-hours/week), (iii) “overweight or obese/active” (≥25 kg/m², ≥8.75 MET-hours/week); and (iv) “overweight or obese/inactive” (≥25 kg/m², <8.75 MET-hours/week) groups at baseline.

Biomarker levels below and above the detection limit were replaced with the mean of the lowest and highest quintile, respectively. Log₂ transformation was applied to adjust for heteroscedasticity in biomarker levels.

Mean values and standard deviations for continuous variables and frequencies and percentages for categorical variables were computed to describe patient characteristics. Multivariate linear regression models were used to assess associations of continuous values of physical activity levels and BMI, as well as combined physical activity and BMI groups with biomarker levels. Stratified analyses were conducted to identify effect modification by sex (female/male), regular aspirin and/or NSAID drug use (yes/no, at least 1 time/week during the preceding month), and tumor site (colon/rectum). If no effect modification by these factors was observed, they were considered potential confounding factors. Models were adjusted for potential confounders, including sex (female/male), age, race (White, non-White), stage at diagnosis (I, II, III, IV—before receipt of any neoadjuvant treatment), cancer site (colon/rectum), neoadjuvant treatment (yes/no), smoking (never, former, current), regular aspirin, and/or NSAID drug use (yes/no, at least 1 time/week during the preceding month), biomarker measurement batch, and study site (HCI, FHCC, and HD). Correlations between independent variables were computed to detect multicollinearity. Robustness of the model and confounding effects of relevant factors were assessed using standard methods including 10% rule and likelihood ratio test. The final models included age, sex, stage at diagnosis, and neoadjuvant treatment. False discovery rate correction was used to account for multiple testing (60). The percent relative difference (RD) was calculated using the β-coefficient derived from the regression models as follows: (exp(β-coefficient) − 1) × 100. Sensitivity analyses were performed excluding patients (i) who received neoadjuvant chemotherapy (n = 185), (ii) who were diagnosed with stage IV colorectal cancer (n = 114), and (iii) who took aspirin or NSAIDs 24 hours prior to the blood draw. All statistical analyses were performed in SAS (version 9.4).

The data generated in this study are not publicly available due to information that could compromise patient privacy or consent but may be available upon reasonable request from the ColoCare investigator team (email: colocarestudy_admin@hci.utah.edu).

Table 1 summarizes the demographic and clinical characteristics of the study population. Patients were on average 62 years old and 60% were male. Most patients were diagnosed with stage II (26%) or stage III (37%) colorectal cancer, and 51% of tumors were located in the rectum. Patients reported average physical activity levels of 12.5 MET-hours/week, but 69% reported to be “inactive.” Most patients were “overweight” or “obese” (64%, n = 372) and had an average BMI of 27.6 kg/m². A small proportion of patients were current smokers (13%), whereas 36% were former and 34% were never smokers. Two thirds of the study population (66%) reported the use of aspirin and/or NSAIDs on a regular basis (at least 1 time/week during the preceding month). Distributions within the combined physical activity and BMI groups were as follows: n = 102 (17%) “normal weight/active,” n = 105 (18%) “normal weight/inactive,” n = 137 (24%) “overweight or obese/active,” n = 235 (41%) “overweight or obese/inactive.” Mean biomarker levels by individual and combined groups of physical activity and BMI are presented in Table 2.

All measured biomarker levels were elevated among “inactive” compared with “active” patients (Table 3). Despite not being statistically significant, CRP and IL6 levels tended to be higher among “inactive” compared with “active” patients [percent relative difference (RD)_CRP = +39%, P = 0.15; RD_IL6 = +36%, P = 0.10]. Similar observations were made when using four physical activity categories indicating a dose–response relationship between physical activity and biomarker levels (Supplementary Table S1). Stronger associations were observed for CRP among nonregular aspirin/NSAID users (Supplementary Table S2). In contrast, differences in SAA levels were more prominent among regular users. However, effect modification by regular aspirin/NSAID use was not statistically significant. No differences in the results were observed when stratifying by sex and tumor site. Similarly, no meaningful differences in the results were observed when excluding patients who received neoadjuvant treatment or who were diagnosed with stage IV disease.

All biomarker levels were elevated among “obese” compared with “normal weight” patients (Table 3). After adjusting for potential confounders, levels of CRP and TNFα statistically significantly differed among “obese” compared with “normal weight” patients (RD_CRP = +88%, P = 0.03; RD_TNFα = +17%, P = 0.02). Stronger associations in “obese” patients were observed for CRP among regular aspirin/NSAID users (Supplementary Table S2). In contrast, differences in IL6 and IL8 levels in “obese” patients and SAA in “overweight” patients were more prominent among nonregular users. Effect modification by regular aspirin/NSAID use was not statistically significant. No differences in the results were observed when stratifying by sex and tumor site. No differences in the results were observed when excluding patients who received neoadjuvant treatment or who were diagnosed with stage IV disease.

We observed differences in biomarker levels when comparing “normal weight/active” patients with other combined physical activity and BMI groups (Table 4). All biomarker levels were elevated among “normal weight/inactive,” “overweight or obese/active,” and “inactive” compared with “normal weight/active” patients (Table 4). CRP levels were statistically significantly and marginally significantly elevated in “overweight or obese/inactive” and “normal weight/inactive” patients, respectively (RD = +103%, P = 0.03; RD = +88%, P = 0.09). In addition, “normal weight/inactive” patients had marginally significantly increased levels of SAA (RD = +67%, P = 0.15). “Overweight or obese/inactive” patients tended to have increased levels of IL6 (RD = +48%, P = 0.15).

When comparing biomarker levels by physical activity group among “overweight/obese” patients, elevated levels across the biomarker panel were observed among “overweight or obese/inactive” patients (Table 5). TNFα levels were statistically significantly higher in “overweight or obese/inactive” patients compared with their “active” counterparts (RD = +16%, P = 0.02). Meaningful differences were further observed for CRP, and IL6 levels (RD_CRP = +43%, P = 0.17; RD_IL6 = +39%, P = 0.16).

There was some evidence that associations for “overweight or obese/inactive” patients were stronger among regular aspirin/NSAID users (Supplementary Tables S3 and S4). However, effect modification was not statistically significant. No differences in the results were observed when either stratifying by sex and tumor site or excluding patients who received neoadjuvant treatment, or who were diagnosed with stage IV disease.

In this large international cohort of colorectal cancer patients, we investigated associations of physical activity and BMI with proinflammatory biomarkers. We observed that obesity was associated with significantly different levels of CRP and TNFα levels compared with “normal weight” patients. Moreover, obese patients reporting physical activity levels below the current guidelines (<8.75 MET-hours/week) had statistically significantly higher CRP levels and marginally significantly TNFα levels as compared with their “active/normal weight” counterpart. We further observed some differences in effect sizes by regular aspirin/NSAID use.

Although not statistically significant, we observed that CRP and IL6 levels were 36% and 39% higher in “inactive” compared with “active” patients. Our data complement previous results from other cross-sectional and randomized controlled trials suggesting an inverse association between physical activity and proinflammatory biomarker levels (11–22). Overall, studies observed similar effect sizes indicating reduced levels of CRP by ∼30%, and IL6 ∼20% among active compared with inactive individuals (11–20). For example, a recently published randomized controlled trial tested the effect of a 12-week exercise program on changes in biomarker levels among 139 primarily breast and a small proportion (36%) of colorectal cancer patients (11). Brown and colleagues observed that exercise alone and in combination with metformin reduced systemic levels of CRP and IL6 by about 30% over the 12-week program period (11). The authors did not observe effect modification by BMI (11). Overall, our results complement existing data in the largest sample of colorectal cancer patients, to date, showing that physical activity may improve systemic inflammatory biomarker profiles in colorectal cancer patients.

Obesity was associated with altered biomarker levels (CRP and TNFα) in comparison with “normal weight” patients. Obesity has been strongly associated with cancer risk including colorectal cancer (1, 3, 4, 8). Chronic low-grade systemic inflammation caused by hypertrophic adipocytes secreting proinflammatory cytokines and adipokines is one of the key hypothesized mechanisms (3, 7, 9). Cross-sectional studies have demonstrated higher levels of proinflammatory biomarkers including CRP and TNFα in obese individuals including colorectal cancer patients (42–45). In addition, weight-loss interventions reduce proinflammatory biomarker levels (e.g., CRP and IL6) among obese individuals including cancer patients (18, 46). We have further previously identified a direct association between visceral adiposity and systemic biomarkers of inflammation and angiogenesis (36). Our study supports the existing body of literature showing BMI-defined obesity is associated with increased proinflammatory biomarker levels among colorectal cancer patients, which may lead to worse clinical outcomes.

Adipose dysfunction and its induced systemic inflammation may not only be found in obese individuals defined based on their BMI. As such, increased physical activity among obese individuals is one hypothesized mechanism underlying the “metabolically healthy obese” phenomenon, which refers to a proportion of obese individuals that do not represent the obesity-characterized phenotype of chronic systemic inflammation and insulin resistance (24, 25). In contrast, normal-weight individuals with insulin resistance and increased inflammation have also been discovered (61–63). Systemic levels of proinflammatory biomarkers, statistically significantly CRP and marginally significantly SAA and IL6 were increased among “inactive” patients regardless of their BMI. In contrast, biomarker levels except for CRP were similar when comparing “normal weight/active” patients with “overweight or obese/active” patients. Other studies testing associations between physical activity and biomarkers yielded inconclusive results showing no changes or an attenuated association with increasing BMI (11, 12). Taken together, we observed associations between combined physical activity and BMI groups and inflammatory biomarkers. Future studies should expand on our results by investigating this association beyond BMI assessments using more comprehensive body composition measurements including computed tomography or dual-energy X-ray (DXA) scans to differentiate and quantity adipose tissue compartments and muscle mass.

This study has several strengths and limitations. To date, this is the largest study investigating systemic inflammation as potential underlying mechanism of the association between energy balance components and colorectal cancer. It is further the first study to test combined associations of physical activity and BMI on systemic inflammation in colorectal cancer patients. Inflammation-related biomarkers were measured following standardized protocols for biospecimen collection, processing, storage, quality controls, and analyses. Overall, the baseline characteristics of our study population are consistent with those of cancer registries (64). The slightly higher proportion of rectal cancer cases (49%) as compared with the general cancer population may be a result of recruitment largely at national Comprehensive Cancer Centers and University clinics, which are more likely to conduct complex surgeries. Drift correction was used to account for batch effects across study sites, and false discovery rate correction was applied to account for multiple testing at the analyses stage. Environmental factors including anti-inflammatory drug use, smoking, and cancer treatment that may influence biomarker levels were accounted for in statistical analyses. Although the ColoCare Study assesses a multitude of potential confounders, the possibility of residual confounding remains. Physical activity was self-reported, which may have introduced reporting bias. Misclassification resulting from reporting bias would be assumed to be nondifferential and, therefore, our results are expected to represent a smaller observed effect size compared with the actual effect size. Blood samples were stored on average for 3–5 years before the biomarker analyses were conducted, which may have introduced measurement errors. As previously noted, prior studies reported robust stability of the analyzed biomarkers over long-term storage (29–35). Longitudinal blood sample collections will be useful in future studies to account for intrapersonal variability of the biomarker levels.

Taken together, we provide evidence of associations between individual and combined physical activity and BMI groups with inflammation-related biomarkers in the largest study of colorectal cancer patients to date. We identified proinflammatory biomarkers as potential interventional targets to improve colorectal cancer survivorship. Effect sizes by different exercise types and intensities have yet to be clarified. Our study suggests that higher physical activity may be a critical lifestyle change that reduces obesity-induced inflammation among cancer patients. To fully elucidate the question of whether or not physical activity can counteract obesity-induced inflammation, exercise intervention studies in overweight or obese cancer patients are needed.

C.A. Warby reports grants from NCI, University of Utah, Stiftung LebensBlicke, and Huntsman Cancer Institute CCPS, grants and nonfinancial support from Huntsman Cancer Foundation during the conduct of the study. B. Gigic reports grants from BMBF and NIH during the conduct of the study. S.A. Cohen reports personal fees from Natera, Kallyope, Istari Oncology, Pfizer, Taiho, Delcath, Bayer, and Isofol outside the submitted work. E.M. Siegel reports grants from NIH NCI during the conduct of the study. No disclosures were reported by the other authors.

C. Himbert: Conceptualization, data curation, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–original draft. C.A. Warby: Data curation, formal analysis, methodology, writing–review and editing. B. Gigic: Data curation, funding acquisition, investigation, writing–review and editing. J. Ose: Conceptualization, data curation, formal analysis, supervision, funding acquisition, writing–review and editing. T. Lin: Formal analysis, validation, investigation, writing–review and editing. R. Viskochil: Data curation, funding acquisition, writing–review and editing. A.R. Peoples: Data curation, supervision, funding acquisition, writing–review and editing. A. Ashworth: Data curation, writing–review and editing. P. Schrotz-King: Data curation, writing–review and editing. C.L. Scaife: Data curation, writing–review and editing. J.N. Cohan: Data curation, writing–review and editing. J. Jedrzkiewicz: Data curation, writing–review and editing. P. Schirmacher: Data curation, funding acquisition, writing–review and editing. W.M. Grady: Data curation, funding acquisition, writing–review and editing. S.A. Cohen: Data curation, writing–review and editing. M. Krane: Data curation, writing–review and editing. J.C. Figueiredo: Data curation, supervision, funding acquisition, investigation, project administration, writing–review and editing. A.T. Toriola: Data curation, supervision, funding acquisition, project administration, writing–review and editing. E.M. Siegel: Data curation, supervision, funding acquisition, writing–review and editing. D. Shibata: Data curation, supervision, funding acquisition, project administration, writing–review and editing. J.L. Round: Conceptualization, supervision, methodology, writing–review and editing. L.C. Huang: Conceptualization, data curation, supervision, investigation, writing–review and editing. C.I. Li: Supervision, funding acquisition, writing–review and editing. M. Schneider: Data curation, supervision, funding acquisition, project administration, writing–review and editing. A. Ulrich: Data curation, supervision, funding acquisition, project administration, writing–review and editing. S. Hardikar: Conceptualization, formal analysis, supervision, funding acquisition, validation, investigation, writing–review and editing. C.M. Ulrich: Conceptualization, resources, data curation, supervision, funding acquisition, validation, investigation, methodology, writing–original draft, writing–review and editing.

C.M. Ulrich, E.M. Siegel, J.C. Figueiredo, D. Shibata, C.I. Li, A.T. Toriola, and M. Schneider have been awarded U01 CA206110. C.M. Ulrich has been awarded R01 CA254108 and R01 CA211705, as well as funding from the German Consortium of Translational Cancer Research (DKTK), German Cancer Research Center, Matthias Lackas Stiftung, ERA-NET, JTC 2012 call on Translational Cancer Research (TRANSCAN), and Federal Ministry of Education and Research (BMBF), Germany, projects 01KT1503 and 01KD2101D. C.M. Ulrich and C.I. Li have been awarded R01 CA189184 and R01 CA207371. S. Hardikar was awarded K07 CA222060. J.N. Cohan was awarded R03 AG067994. C. Himbert was awarded funding from the Stiftung LebensBlicke and Claussen Simon Stiftung. C.M. Ulrich, C Himbert, J. Ose, and S. Hardikar have been awarded funding from the Huntsman Cancer Foundation and Cancer Control and Population Health Sciences (CCPS) at the University of Utah. This study was in addition supported by KL2TR002539 and funding from the Immunology, Inflammation, and Infectious Disease Initiative at the University of Utah.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.