Obesity and metabolic dysfunction are implicated in colorectal cancer development. Appetite-regulating gut hormones might have a role in colorectal cancer risk. We investigated whether circulating levels of the gut hormones ghrelin (analyzed as acyl ghrelin) and Peptide YY (PYY) were associated with subsequent colorectal cancer risk, including clinical and molecular tumor subtypes. We also provide descriptive data on these hormones in relation to background participant characteristics and metabolic biomarkers. This population-based study included 1,010 matched case–control pairs with a median of 12.3 years of follow-up. Acyl ghrelin and PYY were measured by multiplex immunoassay. Data on KRAS and BRAF mutations and microsatellite instability (MSI) status were available for 704 and 708 cases, respectively. Conditional logistic regression models estimated association to colorectal cancer risk. Partial correlation and linear regression were used to investigate relationships between background and metabolic variables and variation in plasma gut hormone concentrations. Acyl ghrelin was not clearly associated with colorectal cancer risk (multivariable OR per 1 SD increase: 1.11; 95% CI, 1.00–1.23). Positive associations were observed for specific subtypes, in particular BRAF-mutated colorectal cancer and right-sided colon cancer, although with nonsignificant heterogeneity. PYY was not related to colorectal cancer risk (multivariable OR per 1 SD: 1.04; 95% CI, 0.95–1.14) or any tumor subtype. In the control participants, ghrelin was inversely correlated with BMI, and PYY was positively correlated with C-peptide and insulin levels. These findings provide limited support for a possible role for ghrelin in colorectal cancer development, primarily in specific anatomical and molecular tumor subtypes.

Prevention Relevance:

The findings of this study do not support a major role for the metabolic gut hormones ghrelin and PYY in colorectal cancer development but suggest the possibility of an involvement for ghrelin in specific tumor subtypes. Elucidating subtype-specific risk factors and mechanisms of carcinogenesis may have implications for precision prevention.

Obesity and excess body fat rank among the most well-established lifestyle-related risk factors for colorectal cancer (1, 2). Plausible biological mechanisms for an independent role of adiposity include insulin resistance, inflammation, and alterations in adipokines and gut microbiome (2). However, metabolic markers related to these potential mechanisms, including growth factors, adipocyte-derived cytokines, and markers of insulin resistance, have shown inconsistent associations with colorectal cancer risk in studies from prospective cohorts, using pre-diagnostic samples (3–6). To help elucidate any underlying etiologic mechanisms, novel biomarkers of energy metabolism and metabolic functions related to body size are of interest. Biomarkers reflecting biological mechanisms might also have a potential application for risk stratification for precision screening or as targets for precision pharmacoprevention.

Probably the best-studied gut hormone is the appetite-regulating hormone leptin, highly associated with obesity and also secreted by adipose tissue (7). High circulating levels of leptin have been suggested to play an etiologic role in carcinogenesis, including colorectal cancer development (8). However, leptin is not established as a single mediator between obesity and colorectal cancer risk (5, 9). Ghrelin, an appetite-regulating gut hormone with metabolic functions, is secreted pre-prandially with decreasing levels shortly after a meal (10), and peptide YY (PYY), a satiety protein secreted from enteroendocrine cells in response to nutrient intake (11) are other less investigated gut hormones, both in general and in relation to colorectal cancer risk specifically. Ghrelin may have a proliferative effect on colorectal cancer cells (12–14) but findings for circulating ghrelin concentrations in relation to subsequent colorectal cancer risk have been mixed (15–17). PYY has, to our knowledge, not been investigated in a prospective setting.

The main objective of this study was to determine whether plasma concentrations of the gut hormones acyl ghrelin (the biologically active form of ghrelin) and PYY were associated with subsequent risk of developing colorectal cancer, including molecular and clinical tumor subtypes. Secondary objectives included assessment of the relationship between acyl ghrelin, PYY and background variables among the control participants, as well as between plasma acyl ghrelin and total ghrelin concentrations in a subset of participants.

Study population

The study participants were selected from two prospective population-based cohorts, the Västerbotten Intervention Programme (VIP, 91.8% of participants) and the Monitoring Trends and Determinants in Cardiovascular Disease Study (MONICA, 8.2% of participants). Both cohorts were initiated in the mid-1980s and are part of the Northern Sweden Health and Disease Study (NSHDS), previously described in more detail (5, 18). The NSHDS currently has approximately 135,000 participants and 240,000 sampling occasions, including >50,000 participants with repeated measurements, and up to 35 years of follow-up (19–21). In VIP, all residents in Västerbotten are invited to a health examination at their primary health care center when they turn 40, 50, and 60 years old (until 1996, also 30-year-olds were invited). In the MONICA study, participants of Västerbotten and Norrbotten ages 25 to 74 years are randomly invited every 4 to 5 years. VIP and MONICA follow very similar protocols, and in both cohorts, the vast majority of blood samples are collected after >8 hours of fasting, then aliquoted and frozen within an hour. Long-term storage is at a central location in −80° freezers. Data collected at the health examination also include extensive lifestyle and health questionnaires, anthropometry measurements, measurements of blood fats and blood pressure, and an oral glucose tolerance test.

Incident colorectal cancer cases diagnosed after cohort participation and prior to May 31, 2016, were identified by linkage to Swedish national registries as described previously (22). This study included 1010 cases with a verified primary colorectal cancer diagnosis and a prediagnostic plasma sample available, as well as 1:1 control participant matched by sex, age, year of sampling, fasting status and, for the large majority (89.5%) of samples, number of freeze–thaw cycles (Fig 1). Participants with a previous cancer diagnosis, other than nonmelanoma skin cancer, were excluded.

Figure 1.

Selection of study participants. The flowchart provides an overview of the study design and presents the inclusions and exclusions for each set of participants and samples in the main and subanalyses. LOD, limit of detection; MI, multiple imputation. MONICA, Monitoring Trends and Determinants in Cardiovascular Disease Study; MSI, microsatellite instability; VIP, Västerbotten Intervention Programme.

Figure 1.

Selection of study participants. The flowchart provides an overview of the study design and presents the inclusions and exclusions for each set of participants and samples in the main and subanalyses. LOD, limit of detection; MI, multiple imputation. MONICA, Monitoring Trends and Determinants in Cardiovascular Disease Study; MSI, microsatellite instability; VIP, Västerbotten Intervention Programme.

Close modal

For all 1,010 case–control pairs, baseline measurements of plasma acyl ghrelin and PYY were available (Fig. 1). Of those, 259 case–control pairs also had a second measurement collected at least 5 years, and generally 10 years, after the baseline measure. In addition, a subset of the 518 participants with repeated measurements also had data on plasma total ghrelin concentrations available from an earlier study (n = 119) (16). The data subsets are illustrated in Fig. 1.

Ethics statement

The study protocol for the present investigation was approved by the Regional Ethical Review Board at Umeå University. All participants provided a written informed consent to use their blood samples for research purposes. All data handling complies with the European Union General Data Protection Regulation. The study conforms with the Code of Ethics of the World Medical Association (Declaration of Helsinki), printed in the British Medical Journal (July 18, 1964).

Plasma and tumor analyses

Levels of acyl ghrelin and total PYY were measured in EDTA plasma using a custom-designed multiplex immunoassay from mesoscale discovery (MSD). The assay was run according to the manufacturer's instructions, as described previously (5). Two aliquots of a pooled plasma control sample were included on each 96-well plate for calculation of coefficients of variation (CV). Inter- and intraassay CVs were, for acyl ghrelin 5.0% and 1.7% respectively, and for PYY 2.9% and 0.9% respectively. In the subset of participants with data on plasma total ghrelin, concentrations were analyzed using sandwich ELISA (Merck). Inter- and intraassay CVs for total ghrelin were 7.9% and 3.1%, respectively. For all plasma analyses, samples were sorted by matched case–control sets, with random placement of the samples within each set.

Colorectal cancer cases with available archival formalin-fixed, paraffin-embedded tumor tissue (n = 841, 83%) were analyzed for BRAFV600E and KRAS (codon 12 and 13) mutations and microsatellite instability (MSI) status. As described previously, KRAS mutations in codons 12 and 13 were detected by Sanger sequencing, BRAF V600E mutations were detected by TaqMan allelic discrimination or digital droplet PCR, and MSI status was determined by IHC or by a PCR-based method (22). As KRAS and BRAF are generally mutually exclusive, cases were classified as either KRAS mutated, BRAF mutated, or KRAS/BRAF wild type. Cases lacking molecular tumor data were generally due to unavailable or insufficient amount of tumor tissue (n = 169), mutations in both KRAS and BRAF (n = 5, considered to be different clones in the same tumor), or inconclusive tumor data (n = 132 for KRAS or BRAF mutation status and n = 133 cases for MSI status, generally due to insufficient amounts of DNA). Of the 836 cases left, 704 had KRAS/BRAF data of whom 24% were KRAS mutated and 22% BRAF mutated, and 708 had MSI data of whom 13% were MSI.

Additional variables

The following potential confounding variables, measured by a health-care professional at cohort participation, were considered; systolic and diastolic blood pressure, fasting plasma glucose, oral glucose tolerance test (2 hours after a 75-g oral glucose load), total cholesterol, triglycerides, and BMI (kg/m2, height and weight measured in light clothes and without shoes). Questionnaire data collected at the time of blood sampling were used for smoking status (current, former, and never smoker), alcohol consumption (abstainers, below-, or above sex-specific median), recreational physical activity on a scale from 1 to 4 based on frequency (1: never; 2: now and then; 3: up to 2 times/week; 4: 3 times/week or more and more vigorous training) and dietary consumption of fiber, whole grain, and red and processed meats in g/day.

Statistical analysis

Missing data for variables at baseline and the repeated sampling occasion were assumed to be missing at random. Therefore, imputation was performed using the fully conditional specification approach (also known as multiple imputation by chained equations, MICE) in SPSS (23). In total, 30 imputed sets were generated in which continuous variables were predicted using predictive mean matching and categorical variables using logistic regression. The imputation models included the exposures ghrelin and PYY, and the covariates age, sex, cohort, sample year, fasting status, smoking status, physical activity, alcohol intake, and BMI. Numbers of missing values for these variables are shown in Table 1.

Table 1.

Plasma concentrations of acyl ghrelin and PYY according to baseline characteristics of cancer-free control participants (n = 1,010).

Acyl ghrelin (pg/mL)PYY (pg/mL)
MedianMedian
n (%)(25–75th percentage)P(25–75th percentage)PMissing, n (%)c,d
All controls 1,010 45.4 (21.8–78.4)  25.9 (16.8–37.7)   
Cohort   0.629a  0.473a 0 (0) 
 VIP 927 (91.8) 45.2 (21.5–78.4)  25.9 (17.0–37.7)   
 MONICA 83 (8.2) 47.5 (23.5–77.1)  26.3 (16.2–34.0)   
Fasting status   0.046b  0.008b 0 (0) 
 0–4 hours 35 (3.5) 38.8 (17.6–51.9)  33.3 (23.2–44.2)   
 4–8 hours 177 (17.5) 54.3 (24.7–84.5)  28.2 (17.7–42.1)   
 >8 hours 798 (79.0) 44.8 (21.1–77.7)  24.9 (16.4–36.6)   
Sex   <0.001a  0.002a 0 (0) 
 Men 525 (52.0) 37.9 (16.3–65.0)  27.2 (18.2–41.6)   
 Women 485 (48.0) 56.6 (28.9–93.7)  24.6 (15.6–34.7)   
Age (years)   <0.001a  0.452a −0 (0) 
 ≤55 504 (49.9) 53.8 (27.4–85.6)  26.5 (16.7–39.2)   
 >55 506 (50.1) 39.2 (17.1–70.6)  25.7 (17.0–36.1)   
BMI (kg/m2  <0.001b  0.073b 10 (1.0) 
 <25 447 (44.2) 55.8 (31.1–98.1)  25.7 (16.4–34.9)   
 25–30 414 (41.0) 39.3 (18.3–72.3)  25.8 (16.6–38.6)   
 >30 139 (13.8) 33.8 (11.5–58.6)  29.4 (19.4–44.1)   
Smoking   0.189b  0.028b 23 (2.3) 
 Nonsmoker 443 (43.9) 46.6 (20.0–74.4)  24.6 (15.8–35.8)   
 Ex-smoker 321 (31.8) 42.9 (21.4–79.8)  27.5 (18.9–37.8)   
 Current smoker 223 (22.0) 53.0 (24.3–84.9)  27.4 (17.2–42.8)   
Physical activitye   0.226b  0.266b 87 (8.6) 
 None 375 (37.1) 41.8 (17.8–81.7)  26.2 (16.6–38.8)   
 Low 248 (24.5) 48.2 (22.4–76.4)  28.4 (18.6–39.0)   
 Moderate 252 (25.0) 51.9 (25.8–79.5)  25.7 (16.4–37.6)   
 High 48 (4.8) 46.6 (27.3–77.3)  23.3 (14.7–32.7)   
Alcohol intakef   <0.001b  0.988b 150 (14.9) 
 None 73 (7.2) 45.5 (20.7–84.7)  27.4 (17.8–34.5)   
 Below median 357 (35.3) 38.1 (17.7–66.9)  25.8 (16.8–37.9)   
 Above median 430 (42.6) 53.6 (24.9–82.2)  25.5 (16.6–37.9)   
Acyl ghrelin (pg/mL)PYY (pg/mL)
MedianMedian
n (%)(25–75th percentage)P(25–75th percentage)PMissing, n (%)c,d
All controls 1,010 45.4 (21.8–78.4)  25.9 (16.8–37.7)   
Cohort   0.629a  0.473a 0 (0) 
 VIP 927 (91.8) 45.2 (21.5–78.4)  25.9 (17.0–37.7)   
 MONICA 83 (8.2) 47.5 (23.5–77.1)  26.3 (16.2–34.0)   
Fasting status   0.046b  0.008b 0 (0) 
 0–4 hours 35 (3.5) 38.8 (17.6–51.9)  33.3 (23.2–44.2)   
 4–8 hours 177 (17.5) 54.3 (24.7–84.5)  28.2 (17.7–42.1)   
 >8 hours 798 (79.0) 44.8 (21.1–77.7)  24.9 (16.4–36.6)   
Sex   <0.001a  0.002a 0 (0) 
 Men 525 (52.0) 37.9 (16.3–65.0)  27.2 (18.2–41.6)   
 Women 485 (48.0) 56.6 (28.9–93.7)  24.6 (15.6–34.7)   
Age (years)   <0.001a  0.452a −0 (0) 
 ≤55 504 (49.9) 53.8 (27.4–85.6)  26.5 (16.7–39.2)   
 >55 506 (50.1) 39.2 (17.1–70.6)  25.7 (17.0–36.1)   
BMI (kg/m2  <0.001b  0.073b 10 (1.0) 
 <25 447 (44.2) 55.8 (31.1–98.1)  25.7 (16.4–34.9)   
 25–30 414 (41.0) 39.3 (18.3–72.3)  25.8 (16.6–38.6)   
 >30 139 (13.8) 33.8 (11.5–58.6)  29.4 (19.4–44.1)   
Smoking   0.189b  0.028b 23 (2.3) 
 Nonsmoker 443 (43.9) 46.6 (20.0–74.4)  24.6 (15.8–35.8)   
 Ex-smoker 321 (31.8) 42.9 (21.4–79.8)  27.5 (18.9–37.8)   
 Current smoker 223 (22.0) 53.0 (24.3–84.9)  27.4 (17.2–42.8)   
Physical activitye   0.226b  0.266b 87 (8.6) 
 None 375 (37.1) 41.8 (17.8–81.7)  26.2 (16.6–38.8)   
 Low 248 (24.5) 48.2 (22.4–76.4)  28.4 (18.6–39.0)   
 Moderate 252 (25.0) 51.9 (25.8–79.5)  25.7 (16.4–37.6)   
 High 48 (4.8) 46.6 (27.3–77.3)  23.3 (14.7–32.7)   
Alcohol intakef   <0.001b  0.988b 150 (14.9) 
 None 73 (7.2) 45.5 (20.7–84.7)  27.4 (17.8–34.5)   
 Below median 357 (35.3) 38.1 (17.7–66.9)  25.8 (16.8–37.9)   
 Above median 430 (42.6) 53.6 (24.9–82.2)  25.5 (16.6–37.9)   

aIndependent two-sample Mann–Whitney U test.

bIndependent two-sample Kruskal–Wallis test.

cMissing category not included in significance tests.

dThe corresponding number and proportion missing of the 1,010 cases were: cohort: 0 (0%), fasting status 0 (0%), sex 0 (0%), age 0 (0%), BMI 10 (1.0%), smoking 20 (2.0%), physical activity 74 (7.3%), and alcohol intake 148 (14.7%).

eOn the basis of self-reported recreational physical activity on a scale from 1 to 4 based on frequency (1, none; 2, now and then; 3, up to 2 times/week; 4, 3 times/week or more and more vigorous training).

fOn the basis of self-reported food frequency questionnaire, median (among users) in women: 1.81 g/day, and in men: 4.80 g/day. BMI, body mass index; PYY, Peptide YY.

Associations between acyl ghrelin, PYY, and colorectal cancer risk were evaluated using conditional logistic regression to estimate ORs per sex-specific 1 SD increase in biomarker plasma concentration as well as across groups of sex-specific quartiles, based on the distribution of the exposure in the control group. In quartile-based analyses, the median of each quartile was used as a continuous variable in logistic regression models to test for linear trends. None of the potential confounders had a material impact on the ORs for ghrelin or PYY in relation to colorectal cancer risk. Three multivariable logistic regression models, conditioned on the matched case sets, were therefore constructed, comprising a limited set of covariates that we considered to have strong theoretical potential relevance. Model 1 was a crude model only conditioned on matched case–control pairs; Model 2 was adjusted for smoking status, physical activity level, and alcohol consumption; and Model 3 additionally included BMI. BMI was included in the multivariable model in a separate step because although it is, perhaps, the most important potential confounder, the relationship between BMI and gut hormones is not fully understood and we cannot exclude the possibility of a mediating role (i.e., gut hormones impacting on body constitution and thereby affecting colorectal cancer risk). We also tested correcting the continuous ORs from Model 3 for regression dilution bias by multiplying the log OR with the reciprocal of the intraclass correlation coefficient (ICC) (24). ICCs, defined as the proportion of between person-variance to total-variance, were estimated by fitting mixed effects models on the subset of individuals with repeated measures of acyl ghrelin and PYY (25). Mixed effects models included case status and age as fixed factors and participant identification code as a random factor. Nonlinear associations were tested using restricted cubic splines with five knots placed at the 5th, 27.5th, 50th, 72.5th, and 95th percentiles.

To investigate whether associations between levels of gut hormones and risk of colorectal cancer risk differed by BRAF/KRAS-status, MSI-status, anatomical subsite, clinical stage of the disease, and time from blood sampling to diagnosis, we estimated subgroup-specific ORs with conditional logistic regression with each control participant matched to its case. For subgroups of BMI (<25, 25–30, >30), we broke up case–control pairs and used logistic regression models, further adjusted for matching factors (age, sex, cohort, sample year, and fasting status) in Model 1 and additionally for smoking status, physical activity, and alcohol consumption in Model 2. Subgroup heterogeneity was tested using the likelihood ratio test of models with and without interaction terms.

Changes in acyl ghrelin and PYY concentrations over time were analyzed using linear mixed models, in addition to calculating ICCs. Log-transformed gut hormone concentrations were modeled, including participant identification codes and case sets as random factors, and case–control status, time from sampling to case diagnosis, smoking, recreational physical activity, alcohol intake, and BMI as fixed factors. An interaction term between case–control status and time was included to test for intraindividual average differences in biomarker concentrations over time between cases and controls. Associations were tested using regression coefficient t tests with degrees of freedom from Satterthwaite approximation.

Our study included two secondary objectives, one concerning the relationship between acyl ghrelin, PYY, and background variables, and the other concerning the relationship between acyl ghrelin and total ghrelin concentrations in plasma. When assessing how acyl ghrelin and PYY related to the background variables, we used linear regression to estimate the proportion of the variation in plasma acyl ghrelin and PYY concentrations that can be predicted from background variables and other metabolic biomarkers. We also calculated partial correlations (using nonimputed data) between metabolic factors and biomarkers adjusted for age, sex, and BMI with Spearman correlations coefficient. These analyses were all conducted using data solely from the control participants. When comparing plasma acyl ghrelin and total ghrelin concentrations, we used the subset of participants from our previous study with data on total ghrelin (16). Linear regression was used to model the relationship between acyl- and total ghrelin at baseline and at the repeated sampling occasion. To preserve statistical power, these analyses included both cases and controls. However, case–control status did not significantly modify the association between acyl- and total ghrelin and was therefore excluded from the model.

Analyses of baseline characteristics and baseline linear regression models were performed using IBM SPSS Statistics for Windows, Version 27.0 (IBM Corp.). For conditional logistic regression, the “survival” R-package was used and for partial correlation network the “ppcor” R-package was used. Mixed models were fitted using the lme4R-package in R v.3.5.0 (R Foundation for Statistical Computing). All statistical tests for significance were two-sided and a P value of below 0.05 was considered statistically significant.

Data availability statement

The data generated in this study are not publicly available due to Swedish Authority for Privacy Protection regulations (the national supervisory authority under the European General Data Protection Regulation, GDPR). Data may be available upon reasonable request to the corresponding author.

Participant characteristics

Median and 25th and 75th percentile plasma concentrations of acyl ghrelin and PYY of the 1,010 control participants at baseline are presented according to background variables in Table 1. Levels of acyl ghrelin were higher in women than in men and in participants ≤55 compared with >55 years of age, with a BMI <25 kg/m2 compared with participants with higher BMIs, and with alcohol intakes above the median compared with below (all P < 0.001). PYY levels were higher in men than women (P = 0.002) and lower in never compared with ever smokers (P = 0.028). Characteristics of cases and controls at baseline have been described in detail elsewhere (5). In brief, the median age for cases and matched controls at baseline was approximately 56.3 years (50.6 and 60.5 in VIP and MONICA participants, respectively), and the median age at diagnosis of cases with colorectal cancer was 66.4 years (66.1 and 72.6 in VIP and MONICA participants, respectively). The median time between sampling and case diagnosis was 12.3 years at the baseline measurement, and 5.8 years at the repeat measurement. Of the participants in this study, 48.0% were women (48.4% and 43.4% in VIP and MONICA, respectively).

Biomarkers and colorectal cancer risk, including molecular and clinical subtypes

Plasma acyl ghrelin and PYY concentrations were not clearly associated with subsequent colorectal cancer risk. Adjusted ORs per 1 sex-specific SD increase in concentration were, for acyl ghrelin 1.11 (95% CI, 1.00–1.23) and for PYY 1.04 (95% CI, 0.95–1.14; Table 2). ORs per quartile of acyl ghrelin and PYY were all nonsignificant as were P for trend (Ptrend > 0.05; Table 2).

Table 2.

Conditional logistic regression presenting the OR and 95% CI for risk of colorectal cancer in all, men, and women per 1 sex-specific SD increase and by sex- specific quartiles (based on the control participants) of acyl ghrelin and peptide YY (PYY) measured in prediagnostic plasma samples.

Acyl ghrelin
Per SD increaseQuartile 1Quartile 2Quartile 3Quartile 4
naOR (95% CI)naRefnaOR (95% CI)naOR (95% CI)naOR (95% CI)PtrendfPhetg
All 1010/1010  255/254  246/252  237/252  272/252   0.97 
Model 1b  1.07 (0.97–1.18)  –  0.97 (0.75–1.26)  0.94 (0.72–1.22)  1.09 (0.83–1.41) 0.51  
Model 2c  1.07 (0.97–1.18)  –  0.94 (0.73–1.22)  0.93 (0.71–1.22)  1.06 (0.81–1.38) 0.60  
Model 3d  1.11 (1.00–1.23)  –  0.96 (0.74–1.25)  0.98 (0.75–1.29)  1.14 (0.86–1.51) 0.27  
Model 3 corre  1.15 (1.00–1.33)           
Men 525/525  130/132  135/131  113/131  147/131    
Model 1b  1.08 (0.94–1.23)  –  1.06 (0.73–1.53)  0.88 (0.60–1.29)  1.16 (0.79–1.69) 0.51  
Model 2c  1.07 (0.93–1.23)  –  1.07 (0.73–1.56)  0.88 (0.60–1.31)  1.12 (0.76–1.66) 0.65  
Model 3d  1.08 (0.93–1.25)  –  1.07 (0.73–1.56)  0.89 (0.60–1.33)  1.14 (0.77–1.69) 0.59  
Model 3 corre  1.11 (0.91–1.36)           
Women 485/485  125/122  111/121  124/120  125/121    
Model 1b  1.06 (0.92–1.22)  –  0.89 (0.62–1.28)  1.00 (0.70–1.44)  1.02 (0.70–1.47) 0.78  
Model 2c  1.07 (0.93–1.23)  –  0.87 (0.60–1.25)  1.03 (0.71–1.49)  1.02 (0.70–1.49) 0.71  
Model 3d  1.14 (0.98–1.33)  –  0.90 (0.62–1.32)  1.13 (0.77–1.64)  1.19 (0.80–1.78) 0.25  
Model 3 corre  1.21 (0.97–1.49)           
Acyl ghrelin
Per SD increaseQuartile 1Quartile 2Quartile 3Quartile 4
naOR (95% CI)naRefnaOR (95% CI)naOR (95% CI)naOR (95% CI)PtrendfPhetg
All 1010/1010  255/254  246/252  237/252  272/252   0.97 
Model 1b  1.07 (0.97–1.18)  –  0.97 (0.75–1.26)  0.94 (0.72–1.22)  1.09 (0.83–1.41) 0.51  
Model 2c  1.07 (0.97–1.18)  –  0.94 (0.73–1.22)  0.93 (0.71–1.22)  1.06 (0.81–1.38) 0.60  
Model 3d  1.11 (1.00–1.23)  –  0.96 (0.74–1.25)  0.98 (0.75–1.29)  1.14 (0.86–1.51) 0.27  
Model 3 corre  1.15 (1.00–1.33)           
Men 525/525  130/132  135/131  113/131  147/131    
Model 1b  1.08 (0.94–1.23)  –  1.06 (0.73–1.53)  0.88 (0.60–1.29)  1.16 (0.79–1.69) 0.51  
Model 2c  1.07 (0.93–1.23)  –  1.07 (0.73–1.56)  0.88 (0.60–1.31)  1.12 (0.76–1.66) 0.65  
Model 3d  1.08 (0.93–1.25)  –  1.07 (0.73–1.56)  0.89 (0.60–1.33)  1.14 (0.77–1.69) 0.59  
Model 3 corre  1.11 (0.91–1.36)           
Women 485/485  125/122  111/121  124/120  125/121    
Model 1b  1.06 (0.92–1.22)  –  0.89 (0.62–1.28)  1.00 (0.70–1.44)  1.02 (0.70–1.47) 0.78  
Model 2c  1.07 (0.93–1.23)  –  0.87 (0.60–1.25)  1.03 (0.71–1.49)  1.02 (0.70–1.49) 0.71  
Model 3d  1.14 (0.98–1.33)  –  0.90 (0.62–1.32)  1.13 (0.77–1.64)  1.19 (0.80–1.78) 0.25  
Model 3 corre  1.21 (0.97–1.49)           
PYY
Per SD increaseQuartile 1Quartile 2Quartile 3Quartile 4
naOR (95% CI)nanaOR (95% CI)naOR (95% CI)naOR (95% CI)PtrendfPhetg
All 1,010/1,010  228/255  243/251  274/252  265/252   0.63 
Model 1b  1.06 (0.97–1.16)  –  1.10 (0.85–1.42)  1.22 (0.95–1.57)  1.19 (0.92–1.54) 0.18  
Model 2c  1.05 (0.96–1.15)  –  1.08 (0.84–1.41)  1.20 (0.93–1.54)  1.17 (0.90–1.52) 0.23  
Model 3d  1.04 (0.95–1-14)  –  1.08 (0.83–1.40)  1.18 (0.91–1.52)  1.15 (0.88–1.49) 0.30  
Model 3 corre  1.08 (0.91–1.28)           
Men 525/525  120/132  121/131  154/131  130/131    
Model 1b  1.04 (0.92–1.18)  –  1.03 (0.71–1.49)  1.30 (0.91–1.84)  1.10 (0.76–1.59) 0.50  
Model 2c  1.02 (0.90–1.16)  –  1.02 (0.70–1.48)  1.25 (0.88–1.79)  1.05 (0.73–1.52) 0.69  
Model 3d  1.02 (0.90–1.16)  –  1.02 (0.70–1.48)  1.25 (0.87–1.78)  1.05 (0.72–1.52) 0.71  
Model 3 corre  1.04 (0.81–1.33)           
Women 485/485  108/123  122/120  120/121  135/121    
Model 1b  1.08 (0.95–1.22)  –  1.17 (0.81–1.68)  1.14 (0.79–1.63)  1.28 (0.89–1.83) 0.23  
Model 2c  1.07 (0.94–1.22)  –  1.11 (0.77–1.62)  1.08 (0.75–1.56)  1.26 (0.87–1.81) 0.24  
Model 3d  1.06 (0.94–1.21)  –  1.10 (0.75–1.60)  1.06 (0.73–1.54)  1.22 (0.84–1.76) 0.31  
Model 3 corre  1.12 (0.88–1.43)           
PYY
Per SD increaseQuartile 1Quartile 2Quartile 3Quartile 4
naOR (95% CI)nanaOR (95% CI)naOR (95% CI)naOR (95% CI)PtrendfPhetg
All 1,010/1,010  228/255  243/251  274/252  265/252   0.63 
Model 1b  1.06 (0.97–1.16)  –  1.10 (0.85–1.42)  1.22 (0.95–1.57)  1.19 (0.92–1.54) 0.18  
Model 2c  1.05 (0.96–1.15)  –  1.08 (0.84–1.41)  1.20 (0.93–1.54)  1.17 (0.90–1.52) 0.23  
Model 3d  1.04 (0.95–1-14)  –  1.08 (0.83–1.40)  1.18 (0.91–1.52)  1.15 (0.88–1.49) 0.30  
Model 3 corre  1.08 (0.91–1.28)           
Men 525/525  120/132  121/131  154/131  130/131    
Model 1b  1.04 (0.92–1.18)  –  1.03 (0.71–1.49)  1.30 (0.91–1.84)  1.10 (0.76–1.59) 0.50  
Model 2c  1.02 (0.90–1.16)  –  1.02 (0.70–1.48)  1.25 (0.88–1.79)  1.05 (0.73–1.52) 0.69  
Model 3d  1.02 (0.90–1.16)  –  1.02 (0.70–1.48)  1.25 (0.87–1.78)  1.05 (0.72–1.52) 0.71  
Model 3 corre  1.04 (0.81–1.33)           
Women 485/485  108/123  122/120  120/121  135/121    
Model 1b  1.08 (0.95–1.22)  –  1.17 (0.81–1.68)  1.14 (0.79–1.63)  1.28 (0.89–1.83) 0.23  
Model 2c  1.07 (0.94–1.22)  –  1.11 (0.77–1.62)  1.08 (0.75–1.56)  1.26 (0.87–1.81) 0.24  
Model 3d  1.06 (0.94–1.21)  –  1.10 (0.75–1.60)  1.06 (0.73–1.54)  1.22 (0.84–1.76) 0.31  
Model 3 corre  1.12 (0.88–1.43)           

Abbreviations: BMI, body mass index; CI, confidence interval; OR, odds ratio; Ref, reference; SD, standard deviation.

aNumber of cases/controls.

bConditioned on matched case-control pairs (matched on age, sex, cohort, sample year, and fasting status.

cAdditionally adjusted for smoking status, recreational physical activity, and alcohol intake.

dAdditionally adjusted for BMI.

eORs corrected for regression dilution.

fThe median of each quartile was used as a continuous variable in logistic regression models to test for linear trends.

gPheterogeneity between estimates for subgroups calculated from likelihood ratio tests.

In subgroup analyses for acyl ghrelin, shown in Table 3, we found positive associations with the risk of colorectal cancer with BRAF-mutation (OR, 1.40; 95% CI, 1.05–1.87), and MSS (OR, 1.17; 95% CI, 1.01–1.34), as well as right-sided colon cancer (OR, 1.29; 95% CI, 1.05–1.60), stage I and II tumors (OR, 1.17; 95% CI, 1.01–1.36), and in the subgroup of participants with BMI <25 (OR, 1.14; 95% CI, 1.00–1.30). However, none of the associations were significantly different from the other corresponding subgroups (all Pheterogeneity > 0.05, Supplementary Tables S1 and S2). For men and women, and lag-time stratification, associations for acyl ghrelin were similar across subgroups (Table 3; Supplementary Table S1), and for PYY, no subgroup-specific associations were observed (Table 4; Supplementary Table S2).

Table 3.

Conditional logistic regression presenting the OR and 95% CI for clinical and molecular subgroups of colorectal cancer per 1 sex-specific SD increase and by sex-specific quartiles of acyl ghrelin concentrations measured in prediagnostic plasma samples.

Acyl Ghrelin
All (OR per 1 SD increase)Quartile 1Quartile 2Quartile 3Quartile 4 (highest)
na0R (95% CI)bPhetcnaRefna0R (95% CI)bna0R (95% CI)bna0R (95% CI)bPtrenddPhetc
BRAF/KRAS status   0.22          0.84 
BRAF-mutated 156/156 1.40 (1.05–1.87)  39/45 – 36/37 1.10 (0.57–2.13) 37/42 1.07 (0.55–2.11) 44/32 1.85 (0.89–3.83) 0.10  
KRAS-mutated 167/167 1.24 (0.94–1.65)  40/46 – 40/38 1.37 (0.70–2.71) 42/42 1.42 (0.71–2.84) 45/41 1.44 (0.73–2.87) 0.36  
 Wildtype 381/381 1.06 (0.89–1.27)  95/96 – 103/101 0.93 (0.60–1.45) 86/93 0.89 (0.56–1.4) 97/91 1.11 (0.70–1.76) 0.61  
MSI status   0.81          0.80 
 MSI 96/96 1.10 (0.76–1.59)  29/30 – 21/24 1.01 (0.43–2.37) 24/21 1.68 (0.68–4.17) 22/21 1.06 (0.38–2.96) 0.62  
 MSS 612/612 1.17 (1.01–1.34)  140/160 – 162/145 1.31 (0.92–1.86) 137/152 1.10 (0.77–1.57) 173/155 1.38 (0.96–1.98) 0.17  
Tumor location   0.30          0.45 
 Right colon 318/318 1.29 (1.05–1.60)  84/96 – 82/84 1.16 (0.74–1.82) 72/72 1.21 (0.75–1.96) 80/66 1.59 (0.95–2.66) 0.08  
 Left colon 296/296 0.99 (0.81–1.22)  76/63 – 67/74 0.65 (0.40–1.05) 68/86 0.62 (0.37–1.04) 85/73 0.89 (0.53–1.49) 0.95  
 Rectum 388/388 1.08 (0.92–1.25)  91/94 – 96/93 1.16 (0.72–1.85) 94/91 1.17 (0.74–1.87) 107/110 1.13 (0.71–1.81) 0.73  
Tumor stage   0.29          0.34 
 Stage I-II 478/478 1.17 (1.01–1.36)  119/128 – 111/124 0.99 (0.68–1.44) 110/106 1.23 (0.82–1.86) 138/120 1.42 (0.94–2.14) 0.06  
 Stage III-IV 468/468 1.06 (0.91–1.25)  118/113 – 122/114 0.99 (0.66–1.47) 106/125 0.84 (0.57–1.24) 122/116 1.04 (0.69–1.57) 0.90  
Lag time   0.35          0.42 
 <9 years 320/320 1.09 (0.92–1.30)  88/87 – 79/81 0.91 (0.58–1.44) 73/79 0.91 (0.56–1.48) 80/73 1.07 (0.66–1.74) 0.69  
 9–15 years 338/338 1.16 (0.95–1.41)  81/81 – 79/95 0.82 (0.51–1.32) 89/87 1.17 (0.73–1.88) 89/75 1.33 (0.79–2.25) 0.13  
 >15 years 352/352 1.07 (0.89–1.28)  86/86 – 88/76 1.23 (0.78–1.94) 75/86 0.87 (0.54–1.4) 103/104 1.05 (0.66–1.66) 0.89  
BMIe   0.97          0.79 
 <25 kg/m2 392/448 1.14 (1.00–1.30)  67/84 – 70/99 0.85 (0.54–1.33) 110/120 1.12 (0.74–1.71) 145/145 1.28 (0.85–1.93) 0.08  
 25–30 kg/m2 462/423 1.07 (0.92–1.26)  130/119 – 124/112 1.10 (0.69–1.44) 94/100 0.87 (0.59–1.28) 114/92 1.13 (0.77–1.67) 0.61  
 >30 kg/m2 156/139 0.98 (0.67–1.42)  58/51 – 52/41 1.10 (0.61–2.00) 33/32 0.90 (0.47–1.74) 13/15 0.82 (0.33–2.00) 0.59  
Acyl Ghrelin
All (OR per 1 SD increase)Quartile 1Quartile 2Quartile 3Quartile 4 (highest)
na0R (95% CI)bPhetcnaRefna0R (95% CI)bna0R (95% CI)bna0R (95% CI)bPtrenddPhetc
BRAF/KRAS status   0.22          0.84 
BRAF-mutated 156/156 1.40 (1.05–1.87)  39/45 – 36/37 1.10 (0.57–2.13) 37/42 1.07 (0.55–2.11) 44/32 1.85 (0.89–3.83) 0.10  
KRAS-mutated 167/167 1.24 (0.94–1.65)  40/46 – 40/38 1.37 (0.70–2.71) 42/42 1.42 (0.71–2.84) 45/41 1.44 (0.73–2.87) 0.36  
 Wildtype 381/381 1.06 (0.89–1.27)  95/96 – 103/101 0.93 (0.60–1.45) 86/93 0.89 (0.56–1.4) 97/91 1.11 (0.70–1.76) 0.61  
MSI status   0.81          0.80 
 MSI 96/96 1.10 (0.76–1.59)  29/30 – 21/24 1.01 (0.43–2.37) 24/21 1.68 (0.68–4.17) 22/21 1.06 (0.38–2.96) 0.62  
 MSS 612/612 1.17 (1.01–1.34)  140/160 – 162/145 1.31 (0.92–1.86) 137/152 1.10 (0.77–1.57) 173/155 1.38 (0.96–1.98) 0.17  
Tumor location   0.30          0.45 
 Right colon 318/318 1.29 (1.05–1.60)  84/96 – 82/84 1.16 (0.74–1.82) 72/72 1.21 (0.75–1.96) 80/66 1.59 (0.95–2.66) 0.08  
 Left colon 296/296 0.99 (0.81–1.22)  76/63 – 67/74 0.65 (0.40–1.05) 68/86 0.62 (0.37–1.04) 85/73 0.89 (0.53–1.49) 0.95  
 Rectum 388/388 1.08 (0.92–1.25)  91/94 – 96/93 1.16 (0.72–1.85) 94/91 1.17 (0.74–1.87) 107/110 1.13 (0.71–1.81) 0.73  
Tumor stage   0.29          0.34 
 Stage I-II 478/478 1.17 (1.01–1.36)  119/128 – 111/124 0.99 (0.68–1.44) 110/106 1.23 (0.82–1.86) 138/120 1.42 (0.94–2.14) 0.06  
 Stage III-IV 468/468 1.06 (0.91–1.25)  118/113 – 122/114 0.99 (0.66–1.47) 106/125 0.84 (0.57–1.24) 122/116 1.04 (0.69–1.57) 0.90  
Lag time   0.35          0.42 
 <9 years 320/320 1.09 (0.92–1.30)  88/87 – 79/81 0.91 (0.58–1.44) 73/79 0.91 (0.56–1.48) 80/73 1.07 (0.66–1.74) 0.69  
 9–15 years 338/338 1.16 (0.95–1.41)  81/81 – 79/95 0.82 (0.51–1.32) 89/87 1.17 (0.73–1.88) 89/75 1.33 (0.79–2.25) 0.13  
 >15 years 352/352 1.07 (0.89–1.28)  86/86 – 88/76 1.23 (0.78–1.94) 75/86 0.87 (0.54–1.4) 103/104 1.05 (0.66–1.66) 0.89  
BMIe   0.97          0.79 
 <25 kg/m2 392/448 1.14 (1.00–1.30)  67/84 – 70/99 0.85 (0.54–1.33) 110/120 1.12 (0.74–1.71) 145/145 1.28 (0.85–1.93) 0.08  
 25–30 kg/m2 462/423 1.07 (0.92–1.26)  130/119 – 124/112 1.10 (0.69–1.44) 94/100 0.87 (0.59–1.28) 114/92 1.13 (0.77–1.67) 0.61  
 >30 kg/m2 156/139 0.98 (0.67–1.42)  58/51 – 52/41 1.10 (0.61–2.00) 33/32 0.90 (0.47–1.74) 13/15 0.82 (0.33–2.00) 0.59  

Abbreviations: BMI, body mass index; MSI, microsatellite instability; MSS, microsatellite stability; OR, odds ratios; PYY, Peptide YY; ref, reference; SD, standard deviation.

aNumber of cases/controls.

bORs are from multivariable Model 3 conditioned on matched case-control pairs (matched on age, sex, cohort, sample year, and fasting status) and adjusted for smoking status, recreational physical activity, alcohol intake, and BMI. Results for minimally and fully adjusted models (with and without BMI) as well as corrected for regression dilution are shown in Supplementary Table S1.

cPheterogeneity between estimates for subgroups calculated from likelihood ratio tests.

dThe median of each quartile was used as a continuous variable in logistic regression models to test for linear trends.

eLogistic regression adjusted for matching variables age, sex, cohort, sampling year, and fasting status (since matched case–control pairs were split in the subgroup-analysis for BMI) and additionally adjusted for smoking status, recreational physical activity, and alcohol intake.

Table 4.

Conditional logistic regression presenting the OR and 95% CI for clinical and molecular subgroups of colorectal cancer per 1 sex-specific SD increase and by sex-specific quartiles of Peptide YY (PYY) measured in prediagnostic plasma samples.

PYY
All (OR per 1 SD increase)Quartile 1Quartile 2Quartile 3Quartile 4 (highest)
na0R (95% CI)bPhetcnaRefna0R (95% CI)bna0R (95% CI)bna0R (95% CI)bPtrenddPhetc
BRAF/KRAS status   0.43          0.10 
BRAF-mutated 156/156 1.11 (0.88–1.41)  41/35 – 33/49 0.49 (0.24–1.00) 39/37 0.79 (0.40–1.57) 43/35 0.97 (0.51–1.88) 0.50  
KRAS-mutated 167/167 1.07 (0.82–1.41)  27/42 – 44/40 1.92 (0.91–4.06) 56/42 2.27 (1.09–4.74) 40/43 1.71 (0.76–3.83) 0.45  
 Wildtype 381/381 0.98 (0.85–1.11)  89/89 – 96/93 1.02 (0.66–1.56) 101/102 0.95 (0.63–1.44) 95/97 0.91 (0.60–1.39) 0.60  
MSI-status   1.00          0.22 
 MSI 96/96 1.01 (0.76–1.35)  22/22 – 20/30 0.64 (0.23–1.80) 30/20 1.48 (0.57–3.84) 24/24 0.98 (0.37–2.57) 0.62  
 MSS 612/612 1.04 (0.93–1.17)  136/146 – 152/147 1.12 (0.79–1.58) 162/160 1.06 (0.76–1.48) 162/159 1.07 (0.76–1.50) 0.88  
Tumor location   0.55          0.91 
 Right colon 318/318 1.08 (0.91–1.28)  76/83 – 80/88 0.99 (0.60–1.63) 77/74 1.11 (0.69–1.77) 85/73 1.23 (0.76–1.96) 0.32  
 Left colon 296/296 1.08 (0.93–1.26)  63/76 – 64/68 1.19 (0.74–1.90) 88/72 1.48 (0.93–2.38) 81/80 1.27 (0.79–2.05) 0.37  
 Rectum 388/388 0.98 (0.84–1.14)  85/93 – 98/93 1.15 (0.75–1.77) 108/104 1.11 (0.74–1.67) 97/98 1.05 (0.69–1.60) 0.98  
Stage   0.65          0.92 
 I–II 478/478 1.03 (0.91–1.17)  110/116 – 122/128 1.03 (0.70–1.52) 124/120 1.08 (0.74–1.57) 122/114 1.09 (0.74–1.60) 0.65  
 III–IV 468/468 1.08 (0.94–1.23)  104/125 – 104/109 1.12 (0.76–1.65) 132/112 1.32 (0.91–1.93) 128/122 1.23 (0.85–1.79) 0.27  
Lag time   0.58          0.58 
 <9 years 320/320 1.06 (0.90–1.25)  65/74 – 67/76 0.98 (0.62–1.57) 101/79 1.43 (0.91–2.25) 87/91 1.03 (0.65–1.64) 0.75  
 9–15 years 338/338 1.06 (0.92–1.22)  78/84 – 88/89 1.08 (0.69–1.70) 93/94 1.03 (0.66–1.60) 79/71 1.16 (0.73–1.85) 0.56  
 >15 years 352/352 0.98 (0.83–1.16)  85/97 – 88/86 1.15 (0.72–1.83) 80/79 1.11 (0.71–1.74) 99/90 1.21 (0.78–1.88) 0.47  
BMId,e   0.73           
 <25 kg/m2 392/448 0.98 (0.86–1.11)  103/116 – 99/113 0.97 (0.66–1.43) 95/118 0.91 (0.62–1.33) 95/101 1.02 (0.69–1.51) 0.93  
 25–30 kg/m2 462/423 1.08 (0.93–1.25)  104/112 – 109/104 1.11 (0.76–1.63) 133/100 1.41 (0.97–2.07) 116/107 1.17 (0.80–1.71) 0.41  
 >30 kg/m2 156/139 1.18 (0.96–1.47)  21/27 – 35/34 1.63 (0.75–3.57) 46/34 1.92 (0.91–4.11) 54/44 1.77 (0.86–3.68) 0.24  
PYY
All (OR per 1 SD increase)Quartile 1Quartile 2Quartile 3Quartile 4 (highest)
na0R (95% CI)bPhetcnaRefna0R (95% CI)bna0R (95% CI)bna0R (95% CI)bPtrenddPhetc
BRAF/KRAS status   0.43          0.10 
BRAF-mutated 156/156 1.11 (0.88–1.41)  41/35 – 33/49 0.49 (0.24–1.00) 39/37 0.79 (0.40–1.57) 43/35 0.97 (0.51–1.88) 0.50  
KRAS-mutated 167/167 1.07 (0.82–1.41)  27/42 – 44/40 1.92 (0.91–4.06) 56/42 2.27 (1.09–4.74) 40/43 1.71 (0.76–3.83) 0.45  
 Wildtype 381/381 0.98 (0.85–1.11)  89/89 – 96/93 1.02 (0.66–1.56) 101/102 0.95 (0.63–1.44) 95/97 0.91 (0.60–1.39) 0.60  
MSI-status   1.00          0.22 
 MSI 96/96 1.01 (0.76–1.35)  22/22 – 20/30 0.64 (0.23–1.80) 30/20 1.48 (0.57–3.84) 24/24 0.98 (0.37–2.57) 0.62  
 MSS 612/612 1.04 (0.93–1.17)  136/146 – 152/147 1.12 (0.79–1.58) 162/160 1.06 (0.76–1.48) 162/159 1.07 (0.76–1.50) 0.88  
Tumor location   0.55          0.91 
 Right colon 318/318 1.08 (0.91–1.28)  76/83 – 80/88 0.99 (0.60–1.63) 77/74 1.11 (0.69–1.77) 85/73 1.23 (0.76–1.96) 0.32  
 Left colon 296/296 1.08 (0.93–1.26)  63/76 – 64/68 1.19 (0.74–1.90) 88/72 1.48 (0.93–2.38) 81/80 1.27 (0.79–2.05) 0.37  
 Rectum 388/388 0.98 (0.84–1.14)  85/93 – 98/93 1.15 (0.75–1.77) 108/104 1.11 (0.74–1.67) 97/98 1.05 (0.69–1.60) 0.98  
Stage   0.65          0.92 
 I–II 478/478 1.03 (0.91–1.17)  110/116 – 122/128 1.03 (0.70–1.52) 124/120 1.08 (0.74–1.57) 122/114 1.09 (0.74–1.60) 0.65  
 III–IV 468/468 1.08 (0.94–1.23)  104/125 – 104/109 1.12 (0.76–1.65) 132/112 1.32 (0.91–1.93) 128/122 1.23 (0.85–1.79) 0.27  
Lag time   0.58          0.58 
 <9 years 320/320 1.06 (0.90–1.25)  65/74 – 67/76 0.98 (0.62–1.57) 101/79 1.43 (0.91–2.25) 87/91 1.03 (0.65–1.64) 0.75  
 9–15 years 338/338 1.06 (0.92–1.22)  78/84 – 88/89 1.08 (0.69–1.70) 93/94 1.03 (0.66–1.60) 79/71 1.16 (0.73–1.85) 0.56  
 >15 years 352/352 0.98 (0.83–1.16)  85/97 – 88/86 1.15 (0.72–1.83) 80/79 1.11 (0.71–1.74) 99/90 1.21 (0.78–1.88) 0.47  
BMId,e   0.73           
 <25 kg/m2 392/448 0.98 (0.86–1.11)  103/116 – 99/113 0.97 (0.66–1.43) 95/118 0.91 (0.62–1.33) 95/101 1.02 (0.69–1.51) 0.93  
 25–30 kg/m2 462/423 1.08 (0.93–1.25)  104/112 – 109/104 1.11 (0.76–1.63) 133/100 1.41 (0.97–2.07) 116/107 1.17 (0.80–1.71) 0.41  
 >30 kg/m2 156/139 1.18 (0.96–1.47)  21/27 – 35/34 1.63 (0.75–3.57) 46/34 1.92 (0.91–4.11) 54/44 1.77 (0.86–3.68) 0.24  

Abbreviations: BMI, body mass index; MSI, microsatellite instability; MSS, microsatellite stability; OR, odds ratios; PYY, Peptide YY; ref, reference; SD, standard deviation.

aNumber of cases/controls.

bORs are from multivariable Model 3 conditioned on matched case-control pairs (matched on age, sex, cohort, sample year, and fasting status) and adjusted for smoking status, recreational physical activity, alcohol intake, and BMI. Results from minimally and fully adjusted models (with and without BMI) as well as for corrected for regression dilution are shown in Supplementary Table S2.

cPheterogeneity between estimates for subgroups calculated from likelihood ratio tests.

dThe median of each quartile was used as a continuous variable in logistic regression models to test for linear trends.

eLogistic regression adjusted for matching variables age, sex, cohort, sampling year, and fasting status (since matched case control pairs were split in the subgroup-analysis for BMI) and additionally adjusted for smoking status, recreational physical activity, and alcohol intake.

Adjusting for potential confounders in models 2 and 3 had marginal impact on the risk estimates compared with the “crude” model (Supplementary Tables S1 and S2). Correcting for regression dilution generally increased the magnitude of ORs and widened confidence intervals (Table 2; Supplementary Tables S1 and S2). Testing for nonlinear relationships between acyl ghrelin, PYY, and colorectal cancer risk, none of the restricted cubic spline models were significantly different from models consisting of linear terms only (Pnonlinear > 0.05).

The mixed models analyses for acyl ghrelin and PYY involving the 259 case–control pairs with repeated measurements 10 years apart, and adjusted for the same set of potential confounders as the fully adjusted conditional logistic regression model, are presented in Fig. 2. Intraindividual concentrations of both biomarkers were generally stable over time in both cases and controls (Fig. 2A and D, respectively). The reproducibility in terms of ICC was 0.71 (95% CI, 0.66–0.75) for acyl ghrelin and 0.52 (95% CI, 0.45–0.58) for PYY.

Figure 2.

Estimated average time trajectories for log-ghrelin and log-PYY concentrations in 259 colorectal cancer cases and 259 time-matched controls (including 146 male and 113 female case–control pairs). AC, The results for ghrelin overall (A), in men (B), and in women (C). DF, The results for PYY overall (D), in men (E), and in women (F). Linear mixed models were used to estimate marginal effects of time and 95% CIs in 259 matched case–control pairs with matched repeated measurements prior to case diagnosis. The x-axis displays years before colorectal cancer diagnosis, where 0 = year of diagnosis. Imputed data were used in a mixed model including participant and matched case–control pairs as random factors, and time between sample collection and case diagnosis, case–control status, an interaction term between time and case–control status, smoking status, recreational physical activity, alcohol intake, and BMI, which were included as fixed factors.

Figure 2.

Estimated average time trajectories for log-ghrelin and log-PYY concentrations in 259 colorectal cancer cases and 259 time-matched controls (including 146 male and 113 female case–control pairs). AC, The results for ghrelin overall (A), in men (B), and in women (C). DF, The results for PYY overall (D), in men (E), and in women (F). Linear mixed models were used to estimate marginal effects of time and 95% CIs in 259 matched case–control pairs with matched repeated measurements prior to case diagnosis. The x-axis displays years before colorectal cancer diagnosis, where 0 = year of diagnosis. Imputed data were used in a mixed model including participant and matched case–control pairs as random factors, and time between sample collection and case diagnosis, case–control status, an interaction term between time and case–control status, smoking status, recreational physical activity, alcohol intake, and BMI, which were included as fixed factors.

Close modal

Determinants of gut hormone concentrations

To assess the contribution of background variables and other metabolic factors and biomarkers to the variation in plasma acyl ghrelin and PYY concentrations, coefficients of determination (R2) were calculated in the 1,010 control participants (Supplementary Table S3). Corresponding partial correlations focusing on energy metabolism variables are illustrated in a correlation network in Supplementary Fig. S1. BMI explained 6.1% of the variation in acyl ghrelin (R2 = 0.061, multivariable P < 0.001), with a correlation coefficient of r = −0.2. Furthermore, serum triglycerides explained 3.2% (P = 0.024) and insulin explained 3.1% (P = 0.004) of the variation in acyl ghrelin. Age and female sex explained 2.6% and 5.5% of the variation in acyl ghrelin, respectively (both P < 0.001). With respect to PYY, C-peptide explained 4.4% of the variation in plasma PYY concentrations (P = 0.042) with a correlation coefficient of r = 0.3. Although insulin did not contribute to variation in PYY in the multivariable analysis (P = 0.275), a positive correlation was observed between these variables (r = 0.3). Female sex, diabetes, and ex-smoker status were all statistically significant determinants of PYY in the multivariable model (P < 0.05), but each explained less than 1% of the variation in PYY concentrations. Acyl ghrelin and PYY concentrations were weak but statistically significant determinants of each other in the multivariable linear models.

Comparison of plasma acyl ghrelin and total ghrelin concentrations

In the subset of individuals with repeated measurements (n = 518), n = 119 had data for both acyl and total ghrelin after exclusions (n = 364 no total ghrelin, n = 34 below the level of detection for total ghrelin, and n = 1 outlier; Fig. 1). In these plasma samples, acyl ghrelin accounted for on average 10% (SD 6%) of total ghrelin concentrations. In linear regression models, acyl ghrelin explained 63% of the variation in total ghrelin concentrations at the first sampling time point and 72% at the second time point (Supplementary Fig. S2). Slope coefficients ranged from 4.74 to 4.99.

In this nested case–control study of 1,010 matched colorectal cancer case–control pairs from a population-based cohort in northern Sweden, we found no clear associations between prediagnostic plasma concentrations of acyl ghrelin or PYY and subsequent risk of colorectal cancer. There were indications of subtype-specific positive associations for acyl ghrelin, particularly for BRAF-mutated colorectal cancer and right-sided colon cancer. However, there was no evidence for a heterogenous association across subtypes in any significance test. In the secondary descriptive analyses, acyl ghrelin was inversely associated with BMI, and PYY was positively associated with C-peptide. In a subset of the participants, plasma levels of acyl and total ghrelin were strongly associated, with acyl ghrelin explaining a moderate to substantial percentage of the variation in total ghrelin levels.

Interest in a potential role for ghrelin status in determining colorectal cancer risk was raised by the Finnish Alpha-Tocopherol, Beta-Carotene Cancer Prevention (ATBC) Study, using prospectively collected blood samples from male smokers. In that study, a time-dependent association was observed between low levels of circulating total ghrelin concentrations and increased colorectal cancer risk for samples collected within 10 years prior to diagnosis, but a positive association for samples collected >20 years before diagnosis (15). We were not able to replicate these findings in a previous analysis of 60 case–control pairs with matched, repeated prediagnostic samples collected 10 years apart (16). In this study, which included the majority of these case–control pairs and with additional data on acyl ghrelin concentrations in a much larger study population, we found no evidence suggesting heterogeneity of association estimates based on time point of sample collection prior to diagnosis.

The analysis of acyl rather than total ghrelin, selected to allow multiplexing with other circulating biomarkers being analyzed simultaneously, may have contributed to the differences in findings in our study compared with the study from the ATBC cohort (17). Acyl ghrelin is rapidly enzymatically degraded to des-acyl ghrelin at room temperature prior to freezing (26). However, in the subset of plasma samples in this study with data on both acyl and total ghrelin, high coefficients of determination (R2) were observed, and the proportion acyl ghrelin/total ghrelin was on average 10% (SD 6%), as would be expected (17), supporting the utility of acyl ghrelin analyzed in our biobank samples. We used a custom multiplex immunoassay to analyze plasma acyl ghrelin concentrations, and a sandwich ELISA for total ghrelin. Although data directly comparing these methods to the traditional radioimmunoassay are scarce, CVs were low in our analyses, and sandwich ELISA has been reported to perform better than radioimmunoassay in terms of cross-reactivity and specificity (27). Furthermore, although des-acyl ghrelin has previously been considered to be inactive, reports increasingly suggest possible roles in biological processes relevant for carcinogenesis (28). Thus, consideration of the relative proportions, and potentially different effects, of acyl and des-acyl ghrelin in vivo may be necessary to fully elucidate the role of ghrelin in colorectal cancer development.

Preclinical and observational clinical data on the role of ghrelin in cancer have been contradictory, suggested to act as both promotor and inhibitor in various cancers (14, 29, 30), and a limited role as a risk biomarker was raised in a recent review with respect to mixed results in earlier prospective studies (17). In well- and moderately differentiated colorectal cancer tumor samples, enhanced expression of ghrelin has been described (12). In human intestinal epithelial cell lines, the promotion of proliferation was shown to be independent of the ghrelin receptor GHS-R1a (13). Furthermore, ghrelin may act as a mediator in malignant tissue by increasing proliferation and anti-apoptotic activity, and the ghrelin–GHS-R axis may have an autocrine–paracrine role in cancer growth and function (31).

In contrast to ghrelin, PYY has not, to our knowledge, been studied in relation to colorectal cancer risk in a prospective setting. In a small case–control study, circulating PYY concentrations were higher in patients with colorectal and gastric cancer compared with control participants, but statistically significant only for gastric cancer (32). In this study, we observed no association between plasma PYY concentrations and subsequent colorectal cancer risk. In the subset of participants with repeated samples collected approximately 10 years apart, PYY levels appeared to rise in female cases compared with controls during the years approaching diagnosis, which could be consistent with the previous case–control study (32) but similar to that study, our finding was not statistically significant. Lower levels of PYY have been hypothesized to predispose to the development and/or maintenance of obesity, and have been shown to partly mediate the reduced appetite and subsequent weight loss after bariatric surgery (33). However, results have not been entirely consistent (34, 35). Intracolonic PYY concentrations appear to be unaffected in human colorectal carcinoma compared with benign colorectal tissue (36). Furthermore, plasma PYY levels have been reported to be similar in patients with colorectal cancer and healthy controls (32), which does not support a role for PYY in colorectal cancer progression.

As appetite-regulating hormones, relationships between both ghrelin and PYY and colorectal cancer risk could be confounded and/or mediated by body size. We considered body size in two different ways, as a separate addition to the multivariable model (with no material effects on risk estimations) and through subgroup analyses stratified by BMI categories. A positive association was observed between acyl ghrelin and colorectal cancer risk solely in participants with BMI <25 kg/m2, although the test for heterogeneity was not significant. Circulating levels of acyl ghrelin were higher in participants with BMI <25 kg/m2 compared with the overweight (BMI 25–30 kg/m2) and obese (BMI >30 kg/m2) groups in our study, in contrast to the Finnish study of male smokers in which BMI did not vary across quartiles of total serum ghrelin concentrations (15). Given the relative paucity of participants with BMI under 20 or above 35, our results are generalizable to normal and overweight individuals and may not capture associations occurring at more extreme body weights.

A secondary aim of this study was to present population-based descriptive data for ghrelin and PYY in relation to participant characteristics, in the 1,010 cancer-free controls. Ghrelin has been reported to correlate with several metabolic factors including BMI (positive), insulin (negative), high-density lipoprotein (positive) (37), and also to have a role in alcohol addiction (positive) (38). The correlation with BMI was replicated in our data and for ghrelin in relation to insulin, we observed an inverse association in the multivariable linear regression analysis. Interestingly, there was also a clear distinction between lower versus higher alcohol consumers (</> median) for the association to plasma ghrelin levels in univariable regression. For PYY, positive correlations with waist-circumference and body fat percentage have been found in women but not in men (35). We observed a positive association between PYY and BMI in the linear regression analyses, but only in the univariable model. In addition, we report novel positive correlations between PYY and both insulin and C-peptide, and the association with insulin remained in the multivariable linear regression model.

A weakness in this study was the lack of more detailed data on body fatness and fat distribution, such as body fat percentage or waist circumference. Information on regular use of aspirin which has been described to reduce the risk for BRAF and KRAS wild-type colorectal cancer by around 30% (39), was also lacking, as was information on presence of atrophic gastritis in participants, which can decrease ghrelin levels (40). Although molecular tumor variables were available for only 83% of the cases, potentially limiting generalizability of results and increasing the risk of chance findings due to multiple testing, this proportion of missing data is similar to other molecular pathologic epidemiologic studies (4).

Major strengths of this study were the use of prospectively collected exposure data and high-quality blood samples in a population-based setting. The cohorts are highly representative of the background population, reducing the risk of selection bias (41). Blood samples were collected and handled according to strict protocols, generally after at least 8 hours of fasting and aliquoted and frozen within an hour of collection, and the fast processing, efficient multiplex assay (MSD) with high sensitivity yielded low CVs. These aspects ensured high quality not only for the investigation of the gut hormones in relation to colorectal cancer risk, but also for the secondary aim of providing descriptive data on their relation to background characteristics and metabolic biomarkers. The large sample size was another important strength, with 1,010 participants who developed colorectal cancer after blood sampling, and 1,010 matched control participants. In particular, archival tumor tissue was acquired and successfully analyzed for molecular markers in approximately 700 cases. Although subtypes of colorectal cancer can be investigated by other means, such as Consensus Molecular Subtypes (42), the features used in the present study are clinically relevant and are frequently used in molecular pathologic epidemiology studies of colorectal cancer (43, 44). In addition, the long range of follow-up times between blood sampling and diagnosis (median 12.3 years) allowed for time-stratified analyses, to account for the possibility of reverse causation or differential effects on tumor initiation and progression. Finally, we leveraged repeated samples for a subgroup of the study participants for the evaluation of intra-individual stability and reliability of the gut hormones over time and to investigate prediagnostic time trajectories.

In conclusion, in this large, population-based study, plasma concentrations of the gut hormones acyl ghrelin and PYY showed no clear associations with subsequent risk of colorectal cancer. For acyl ghrelin, subtype-specific positive associations were observed, particularly for BRAF-mutated colorectal cancer and right-sided colon cancer, although with nonsignificant heterogeneity. In a subset of participants with repeated measures, approximately 10 years apart, time trajectories were generally similar in cases and controls for both acyl ghrelin and PYY during the years prior to colorectal cancer diagnosis of the cases. Finally, we replicated previous reports of an inverse association between circulating ghrelin concentrations and BMI and made a novel observation of a positive relationship between circulating PYY concentrations and C-peptide and insulin.

R. Myte reports employment with Astra Zeneca stock and ownership of Astra Zeneca stock. R. Palmqvist reports grants from Swedish Cancer Society during the conduct of the study. B. Van Guelpen reports grants from Swedish Cancer Society (CAN 2017/581 and 2014/780), Cancer Research Foundation in Northern Sweden, Umeå University, Lion's Cancer Research Foundation, Umeå University, and Knut and Alice Wallenberg Foundation during the conduct of the study. No disclosures were reported by the other authors.

S. Bodén: Conceptualization, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. J. Harbs: Formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. A. Sundkvist: Conceptualization, investigation, writing–original draft, writing–review and editing. K. Fuchs: Investigation, writing–review and editing. R. Myte: Formal analysis, visualization, methodology, writing–review and editing. B. Gylling: Data curation, investigation, writing–review and editing. C. Zingmark: Investigation, writing–review and editing. A. Löfgren Burström: Investigation, writing–review and editing. R. Palmqvist: Data curation, funding acquisition, investigation, writing–review and editing. S. Harlid: Conceptualization, data curation, supervision, investigation, writing–review and editing. B. Van Guelpen: Conceptualization, data curation, supervision, funding acquisition, methodology, writing–original draft, writing–review and editing.

Acknowledgements to additional funders of this work that did not fall under one of the listed categories: The Faculty of Medicine, Umeå University (to B. Van Guelpen), and the regional agreement between Umeå University and Region Västerbotten (to B. Van Guelpen). We thank all the participants in the VIP and MONICA cohorts of the NSHDS. We thank Robert Johansson, Åsa Ågren, Veronika Hellström, and their colleagues at the Biobank Research Unit and Department of Public Health and Clinical Medicine, Umeå University for exceptional assistance. We also would like to thank the staff at Biobanken Norr, Region Västerbotten, as well as Åsa Stenberg and Roger Stenling at the Department of Medical Biosciences, Pathology, Umeå University, for invaluable assistance with the tumor tissue retrieval and analyses. Swedish Cancer Society (CAN 2017/581 and 2014/780 to B. Van Guelpen, and 2012/0501 to R. Palmqvist). Cancer Research Foundation in Northern Sweden to B. Van Guelpen (AMP 21–1039) and to S. Bodén (AMP 20–1015 and AMP 19–984). Lions Cancer Research Foundation, Umeå University to B. Van Guelpen (LP 22–2307, LP 20–2226, LP 19–2206, and LP 18–2189), and to S. Bodén (LP 18–2175). The Knut and Alice Wallenberg Foundation (B. Van Guelpen).

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.

Note: Supplementary data for this article are available at Cancer Prevention Research Online (http://cancerprevres.aacrjournals.org/).

1.
World Cancer Research Fund/American Institute for Cancer Research
.
Continuous Update Project. The Third Expert Report 2018
[
Internet
].
World Cancer Research Fund International
,
London
,
UK
;
2018
[
cited 2022 Jun 27
]. https://www.wcrf.org/diet-activity-and-cancer/cancer-types/colorectal-cancer/
2.
Keum
N
,
Giovannucci
E
.
Global burden of colorectal cancer: emerging trends, risk factors and prevention strategies
.
Nat Rev Gastroenterol Hepatol
2019
;
16
:
713
32
.
3.
Aleman
JO
,
Eusebi
LH
,
Ricciardiello
L
,
Patidar
K
,
Sanyal
AJ
,
Holt
PR
, et al
.
Mechanisms of obesity-induced gastrointestinal neoplasia
.
Gastroenterology
2014
;
146
:
357
73
.
4.
Inamura
K
,
Song
M
,
Jung
S
,
Nishihara
R
,
Yamauchi
M
,
Lochhead
P
, et al
.
Prediagnosis plasma adiponectin in relation to colorectal cancer risk according to KRAS mutation status
.
J Natl Cancer Inst
2016
;
108
:
djv363
.
5.
Myte
R
,
Harlid
S
,
Sundkvist
A
,
Gylling
B
,
Häggström
J
,
Zingmark
C
, et al
.
A longitudinal study of prediagnostic metabolic biomarkers and the risk of molecular subtypes of colorectal cancer
.
Sci Rep
2020
;
10
:
5336
.
6.
Kabat
GC
,
Kim
MY
,
Stefanick
M
,
Ho
GYF
,
Lane
DS
,
Odegaard
AO
, et al
.
Metabolic obesity phenotypes and risk of colorectal cancer in postmenopausal women
.
Int J Cancer
2018
;
143
:
543
51
.
7.
Klok
MD
,
Jakobsdottir
S
,
Drent
ML
.
The role of leptin and ghrelin in the regulation of food intake and body weight in humans: a review
.
Obes Rev
2007
;
8
:
21
34
.
8.
Modzelewska
P
,
Chludzińska
S
,
Lewko
J
,
Reszeć
J
.
The influence of leptin on the process of carcinogenesis
.
Contemp Oncol (Pozn)
2019
;
23
:
63
8
.
9.
Aleksandrova
K
,
Boeing
H
,
Jenab
M
,
Bueno-de-Mesquita
HB
,
Jansen
E
,
van Duijnhoven
FJ
, et al
.
Leptin and soluble leptin receptor in risk of colorectal cancer in the european prospective investigation into cancer and nutrition cohort
.
Cancer Res
2012
;
72
:
5328
37
.
10.
Muller
TD
,
Nogueiras
R
,
Andermann
ML
,
Andrews
ZB
,
Anker
SD
,
Argente
J
, et al
.
Ghrelin
.
Mol Metab
2015
;
4
:
437
60
.
11.
Holzer
P
,
Reichmann
F
,
Farzi
A
.
Neuropeptide Y, peptide YY and pancreatic polypeptide in the gut-brain axis
.
Neuropeptides
2012
;
46
:
261
74
.
12.
Waseem
T
,
Javaid Ur
R
,
Ahmad
F
,
Azam
M
,
Qureshi
MA
.
Role of ghrelin axis in colorectal cancer: a novel association
.
Peptides
2008
;
29
:
1369
76
.
13.
Waseem
T
,
Duxbury
M
,
Ashley
SW
,
Robinson
MK
.
Ghrelin promotes intestinal epithelial cell proliferation through PI3K/Akt pathway and EGFR trans-activation both converging to ERK 1/2 phosphorylation
.
Peptides
2014
;
52
:
113
21
.
14.
Lien
GS
,
Lin
CH
,
Yang
YL
,
Wu
MS
,
Chen
BC
.
Ghrelin induces colon cancer cell proliferation through the GHS-R, Ras, PI3K, Akt, and mTOR signaling pathways
.
Eur J Pharmacol
2016
;
776
:
124
31
.
15.
Murphy
G
,
Cross
AJ
,
Dawsey
SM
,
Stanczyk
FZ
,
Kamangar
F
,
Weinstein
SJ
, et al
.
Serum ghrelin is associated with risk of colorectal adenocarcinomas in the ATBC study
.
Gut
2018
;
67
:
1646
51
.
16.
Sundkvist
A
,
Myte
R
,
Palmqvist
R
,
Harlid
S
,
Van Guelpen
B
.
Plasma ghrelin is probably not a useful biomarker for risk prediction or early detection of colorectal cancer
.
Gut
2019
;
68
:
373
4
.
17.
Kasprzak
A
.
Role of the ghrelin system in colorectal cancer
.
Int J Mol Sci
2022
;
23
:
5380
.
18.
Bodén
S
,
Myte
R
,
Harbs
J
,
Sundkvist
A
,
Zingmark
C
,
Löfgren Burström
A
, et al
.
C-reactive protein and future risk of clinical and molecular subtypes of colorectal cancer
.
Cancer Epidemiol Biomarkers Prev
2020
;
29
:
1482
91
.
19.
Benckert
M
,
Lilja
M
,
Soderberg
S
,
Eliasson
M
.
Improved metabolic health among the obese in six population surveys 1986 to 2009: the Northern Sweden MONICA study
.
BMC obesity
2015
;
2
:
7
.
20.
Norberg
M
,
Wall
S
,
Boman
K
,
Weinehall
L
.
The vasterbotten intervention programme: background, design and implications
.
Glob Health Action
2010
;
3
. doi: .
21.
Northern Sweden Health and Disease Study, NSHDS
[
Internet
]. https://www.umu.se/forskning/infrastruktur/northern-sweden-health-and-disease-study-nshds/:
Umeå University, Umeå
;
2022
[
cited 2022 Aug 23
].
About the infrastructure for the NSHDS
].
22.
Myte
R
,
Gylling
B
,
Haggstrom
J
,
Haggstrom
C
,
Zingmark
C
,
Lofgren Burstrom
A
, et al
.
Metabolic factors and the risk of colorectal cancer by KRAS and BRAF mutation status
.
Int J Cancer
2019
;
145
:
327
37
.
23.
White
IR
,
Royston
P
,
Wood
AM
.
Multiple imputation using chained equations: issues and guidance for practice
.
Stat Med
2011
;
30
:
377
99
.
24.
Morgan
KE
,
Cook
S
,
Leon
DA
,
Frost
C
.
Reflection on modern methods: calculating a sample size for a repeatability sub-study to correct for measurement error in a single continuous exposure
.
Int J Epidemiol
2019
;
48
:
1721
6
.
25.
Rosner
B
.
Fundamentals of biostatistcs
. 7th ed.
Brooks Cole, Salt Lake City, Utah, 303
:
Duxbury Press
;
1995
.
888 (568–71)
.
26.
Deschaine
SL
,
Leggio
L
.
Understanding plasma treatment effect on human acyl-ghrelin concentrations
.
Eur Rev Med Pharmacol Sci
2020
;
24
:
1585
9
.
27.
Prudom
C
,
Liu
J
,
Patrie
J
,
Gaylinn
BD
,
Foster-Schubert
KE
,
Cummings
DE
, et al
.
Comparison of competitive radioimmunoassays and two-site sandwich assays for the measurement and interpretation of plasma ghrelin levels
.
J Clin Endocrinol Metab
2010
;
95
:
2351
8
.
28.
Au
CC
,
Furness
JB
,
Brown
KA
.
Ghrelin and breast cancer: emerging roles in obesity, estrogen regulation, and cancer
.
Front Oncol
2016
;
6
:
265
.
29.
Lin
TC
,
Hsiao
M
.
Ghrelin and cancer progression
.
Biochim Biophys Acta
2017
;
1868
:
51
7
.
30.
Spiridon
IA
,
Ciobanu
DGA
,
Giușcă
SE
,
Căruntu
ID
.
Ghrelin and its role in gastrointestinal tract tumors (Review)
.
Mol Med Rep
2021
;
24
:
663
.
31.
Nikolopoulos
D
,
Theocharis
S
,
Kouraklis
G
.
Ghrelin's role on gastrointestinal tract cancer
.
Surg Oncol
2010
;
19
:
e2
e10
.
32.
Zygulska
AL
,
Furgala
A
,
Kaszuba-Zwoińska
J
,
Krzemieniecki
K
,
Gil
K
.
Changes in plasma levels of cholecystokinin, neurotensin, VIP and PYY in gastric and colorectal cancer - preliminary results
.
Peptides
2019
;
122
:
170148
.
33.
Stratis
C
,
Alexandrides
T
,
Vagenas
K
,
Kalfarentzos
F
.
Ghrelin and peptide YY levels after a variant of biliopancreatic diversion with Roux-en-Y gastric bypass versus after colectomy: a prospective comparative study
.
Obes Surg
2006
;
16
:
752
8
.
34.
Karra
E
,
Chandarana
K
,
Batterham
RL
.
The role of peptide YY in appetite regulation and obesity
.
J Physiol
2009
;
587
:
19
25
.
35.
Cahill
F
,
Ji
Y
,
Wadden
D
,
Amini
P
,
Randell
E
,
Vasdev
S
, et al
.
The association of serum total peptide YY (PYY) with obesity and body fat measures in the CODING study
.
PLoS One
2014
;
9
:
e95235
.
36.
El-Salhy
M
,
Mazzawi
T
,
Gundersen
D
,
Hatlebakk
JG
,
Hausken
T
.
The role of peptide YY in gastrointestinal diseases and disorders (review)
.
Int J Mol Med
2013
;
31
:
275
82
.
37.
Purnell
JQ
,
Weigle
DS
,
Breen
P
,
Cummings
DE
.
Ghrelin levels correlate with insulin levels, insulin resistance, and high-density lipoprotein cholesterol, but not with gender, menopausal status, or cortisol levels in humans
.
J Clin Endocrinol Metab
2003
;
88
:
5747
52
.
38.
Koopmann
A
,
Bach
P
,
Schuster
R
,
Bumb
JM
,
Vollstädt-Klein
S
,
Reinhard
I
, et al
.
Ghrelin modulates mesolimbic reactivity to alcohol cues in alcohol-addicted subjects: a functional imaging study
.
Addict Biol
2019
;
24
:
1066
76
.
39.
Amitay
EL
,
Carr
PR
,
Jansen
L
,
Walter
V
,
Roth
W
,
Herpel
E
, et al
.
Association of aspirin and nonsteroidal anti-inflammatory drugs with colorectal cancer risk by molecular subtypes
.
J Natl Cancer Inst
2019
;
111
:
475
83
.
40.
Sakao
Y
,
Ohashi
N
,
Sugimoto
M
,
Ichikawa
H
,
Sahara
S
,
Tsuji
T
, et al
.
Gender differences in plasma ghrelin levels in hemodialysis patients
.
Ther Apher Dial
2019
;
23
:
65
72
.
41.
Norberg
M
,
Blomstedt
Y
,
Lönnberg
G
,
Nyström
L
,
Stenlund
H
,
Wall
S
, et al
.
Community participation and sustainability—evidence over 25 years in the västerbotten intervention programme
.
Glob Health Action
2012
;
5
:
1
9
.
42.
Thanki
K
,
Nicholls
ME
,
Gajjar
A
,
Senagore
AJ
,
Qiu
S
,
Szabo
C
, et al
.
Consensus molecular subtypes of colorectal cancer and their clinical implications
.
Int Biol Biomed J
2017
;
3
:
105
11
.
43.
Leggett
B
,
Whitehall
V
.
Role of the serrated pathway in colorectal cancer pathogenesis
.
Gastroenterology
2010
;
138
:
2088
100
.
44.
Ogino
S
,
Chan
AT
,
Fuchs
CS
,
Giovannucci
E
.
Molecular pathological epidemiology of colorectal neoplasia: an emerging transdisciplinary and interdisciplinary field
.
Gut
2011
;
60
:
397
411
.
This open access article is distributed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license.