Abstract
Circulating adiponectin is inversely related to the risk of colorectal cancer. However, its influence on colorectal cancer survival is unclear. We conducted a prospective study to evaluate the association between prediagnostic plasma levels of adiponectin and mortality in patients with colorectal cancer. We identified 621 incident colorectal cancer cases who provided blood specimens prior to diagnosis within the Nurses' Health Study (NHS) and Health Professionals Follow-up Study (HPFS). Cox proportional hazards models were used to calculate HRs and 95% confidence intervals (CI). After a median follow-up of 9 years, there were 269 (43%) total deaths, of which 181 (67%) were due to colorectal cancer. Compared with participants in the lowest quartile of adiponectin, those in the highest quartile had multivariate HRs of 1.89 (95% CI, 1.21–2.97; Ptrend = 0.01) for colorectal cancer–specific mortality and 1.66 (95% CI, 1.15–2.39; Ptrend = 0.009) for overall mortality. The apparent increased risk in colorectal cancer–specific mortality was more pronounced in patients with metastatic disease (HR, 3.02: 95% CI, 1.50–6.08). Among patients with colorectal cancer, prediagnostic plasma adiponectin is associated with an increased risk of colorectal cancer–specific and overall mortality and is more apparent in patients with metastatic disease. Adiponectin may be a marker for cancers which develop through specific pathways that may be associated with worsened prognosis. Further studies are needed to validate these findings. Cancer Prev Res; 8(12); 1138–45. ©2015 AACR.
Introduction
Colorectal cancer is the fourth most common cancer and the second leading cause of cancer-related deaths in the United States (1). Obesity is a well-established risk factor for colorectal cancer (2, 3). Furthermore, prediagnostic obesity has been associated with worsened survival among patients with colorectal cancer (4, 5). However, the underlying mechanism by which obesity influences the development of colorectal cancer or outcomes after colorectal cancer diagnosis is unclear. Several mechanisms have been hypothesized, including alteration of the adipokine milieu, perturbations of insulin-like growth factor 1 (IGF1) axis, and chronic low-grade inflammation (6, 7).
Adipose tissue is the largest endocrine organ responsible for the regulation of inflammation and metabolism and the synthesis of cytokines and hormones (8). The most abundant hormone secreted by adipose tissue is adiponectin, a 30-kDa protein hormone that exists as low-molecular-weight (LMW) and high-molecular-weight (HMW) multimers in the circulation (9). Its multitude of metabolic functions is effected by binding primarily to adiponectin receptor 1 (ADIPOR1) that is found in skeletal muscle and adiponectin receptor 2 (ADIPOR2) that resides in hepatocytes (10). Accumulating evidence supports an inverse relationship between circulating adiponectin and obesity, suggesting the possibility that adiponectin may mediate the biologic link between obesity and colorectal cancer (11, 12). In addition, higher levels of expression of adiponectin receptors are observed in colorectal cancer than in normal tissues, further supporting this hypothesis (13).
To date, the pleiotropic roles of adiponectin in carcinogenesis are complex and remain controversial. Its anticarcinogenic properties have been attributed to the activation of protein kinase, AMP-activated, alpha 2 catalytic subunit (PRKAA2) pathway, leading to apoptosis and antiproliferation (14), direct inhibition of the phosphoinositide 3 kinase (PI3K)/v-akt murine thymoma viral oncogene homolog 1 (AKT1) pathway which is responsible for cell survival (15), and modulation of insulin sensitization (16) and inflammation (17). In contrast, adiponectin has been shown to propagate oncogenesis through stimulation of proinflammatory cytokines such as chemokine (C-X-C motif) ligand 8 (CXCL8) (18), inhibition of apoptosis via activation of PRKAA2/sirtuin 1 (SIRT1)/peroxisome proliferator-activated receptor gamma, coactivator 1 alpha (PPARGC1A) signaling pathway (19), and promotion of angiogenesis (20, 21) and colonic proliferation (18).
Several epidemiologic studies have established an inverse correlation between circulating adiponectin and the incidence of colorectal cancer (22–25). Among 616 incident colorectal cancer cases and 1,205 controls within the Nurses' Health Study (NHS) and Health Professionals Follow-up Study (HPFS), we have reported that plasma adiponectin was significantly associated with a lower risk of colorectal cancer among men [relative risk (RR), 0.55, 95% confidence interval (CI), 0.35–0.86; Ptrend = 0.02] but not women (RR, 0.96; 95% CI, 0.67–1.39; Ptrend = 0.74; ref. 25). In contrast, there is a paucity of data on the impact of adiponectin on survival among patients with established colorectal cancer. The only prospective study showed that adiponectin (measured at time of diagnosis) was not predictive of colorectal cancer survival (26). Another recent study reported that adiponectin (measured at time of diagnosis) conferred a worse prognosis in patients with hepatocellular carcinoma (HCC; ref. 27). Hence, we prospectively assessed the influence of prediagnostic plasma adiponectin on mortality in patients with colorectal cancer.
Materials and Methods
Study population
Our study composed of 621 incident colorectal cancer cases within the NHS and the HPFS cohorts. The NHS began in 1976 and enrolled 121,700 U.S. female registered nurses, aged 30 to 55 years. The HPFS began in 1986 and enrolled 51,529 male health professionals (podiatrists, dentists, osteopathic physicians, veterinarians, pharmacists, and optometrists), aged 40 to 75 years. Biennially, follow-up questionnaires were administered to collect information with respect to lifestyle factors such as smoking, physical activity, body weight, family history of colorectal cancer, use of aspirin and nonsteroidal anti-inflammatory drug (NSAID), endoscopic screening, and medical history. Dietary information was updated every 4 years using validated food frequency questionnaires (FFQ; refs. 28, 29). High follow-up rates were appreciated for both cohorts, 95.4% in NHS and 95.9% in HPFS. The Institutional Review Board at the Brigham and Women's Hospital and the Harvard School of Public Health approved the study. The completion of self-administered questionnaires by participants was considered to imply informed consent.
Exposure assessment
Phlebotomy kits were mailed to all the participants. A total of 32,826 NHS participants and 18,225 HPFS participants returned blood samples on ice packs by overnight courier from 1989 to 1990 and 1993 to 1995, respectively. Upon receipt of the blood samples in the laboratory, they were immediately centrifuged, aliquoted into plasma, and stored in continuously monitored liquid nitrogen freezers (−130 °C or below) until used in immunoassays. More than 95% of the blood samples arrived in our laboratory within 26 hours of phlebotomy. Details regarding blood collection, processing, and storage of plasma aliquots within these two cohorts have been previously described (30, 31).
Identification of study participants
Individuals with incident colorectal cancer through 2010 follow-up were eligible for this study if they provided a prediagnostic blood specimen, completed the baseline questionnaire, and did not have a history of inflammatory bowel disease or other cancer (except nonmelanoma skin cancer) prior to diagnosis of colorectal cancer. We obtained written permission from participants who reported a diagnosis of colorectal cancer to retrieve their medical and pathology reports. Study physicians who were blinded to the exposure data reviewed all the records and verified the tumor location, stage, and histologic subtype. We searched deaths of nonrespondents through the National Death Index (NDI) and further determined whether the deceased participant had a prior diagnosis of colorectal cancer (32, 33). Through this process, we identified 347 colorectal cancer cases in the NHS and 274 colorectal cancer cases in the HPFS cohorts.
Ascertainment of death
We identified deaths through next-of-kin and the NDI. Mortality follow-up was more than 98% complete (32, 33). For all deaths, we sought information to determine the cause through review of death certificates and medical records.
Laboratory analyses
Plasma levels of adiponectin for all the participants were measured using ELISA from ALPCO Diagnostics (34). To assess laboratory precision, quality control samples were interspersed randomly among the case–control samples. This generated an interbatch coefficient of variation of 8.6%. All assays were conducted by personnel blinded to the quality control status and participants' information. In a subset of participants, we also previously measured adiponectin using a separate ELISA from LINCO (31). For these participants, we observed a correlation of 0.79 between the two assay methods. Analysis of the biomarkers was achieved in a single run in HPFS and two runs in NHS. The run-specific cutoff points were used for association analysis in the NHS to account for laboratory variation. Details regarding measurements of other biomarkers including high sensitivity C-reactive protein (CRP), interleukin 6 (IL6), soluble tumor necrosis factor receptor superfamily, member 1B (TNFRSF1B), IGF1, and insulin-like growth factor binding protein 3 (IGFBP3) have been described in our prior studies (35, 36).
Statistical analyses
We categorized plasma adiponectin level into quartiles on the basis of the known distribution of adiponectin levels among a cohort of controls without cancer that were matched to the cases of colorectal cancer in the present study (25).We calculated means (SD), medians (interquartile ranges, IQR), and proportions for baseline characteristics of study participants in each quartile of adiponectin at the time of blood draw. ANOVA for continuous variables and χ2 tests for categorical variables were used to evaluate differences across quartiles. Cox proportional hazards model was used to calculate HR and 95% confidence interval (CI) for colorectal cancer–specific and overall mortality associated with each quartile of adiponectin. We conducted age-adjusted and multivariate analyses, stratified by age and adjusting for other predictors of survival [date of blood draw, cohort (sex), ethnicity, body mass index (BMI) at time of blood draw, physical activity (MET-h/wk) at time of blood draw, current or past smoking, dietary factors (consumption of alcohol, folate, calcium and red meat), family history of colorectal cancer, regular use of aspirin/NSAIDs, stage at diagnosis, grade (poor/unknown vs. well vs. moderate), site of primary cancer (colon vs. rectum), inflammatory markers (CRP, IL6 and soluble TNFRSF1B), and IGF1/IGFBP3 ratio]. Tests for linear trend were conducted using the median value of each quartile as a continuous variable in the regression models. The proportionality hazards assumption for covariates was tested by the Harrell and Lee test (37).
Further analyses were performed according to selected subgroups (age at blood draw, gender, BMI, site of primary tumor, stage at diagnosis, and histologic grade) to determine whether they modified the association of adiponectin and survival. Tests of interactions between adiponectin and potentially modifying covariates were assessed by entering in the models the cross product of adiponectin with the covariate of interest and assessing their significance using the Wald test. All statistical tests were 2-sided and P < 0.05 was assumed for statistical significance. All analyses were performed using SAS v. 9.3 (SAS Institute, Inc.).
Results
Baseline characteristics
Among the 621 eligible participants, there were 269 (43%) total deaths, of which 181 (67%) were due to colorectal cancer. The median follow-up time for participants who were alive through the end of follow-up from date of diagnosis was 9.1 years (IQR, 6.1–12.1). Plasma collection was performed at a median of 9.8 years (SD, 5.0 years) for NHS and 6.3 years (SD, 3.6 years) for HPFS before diagnosis of colorectal cancer. Baseline characteristics according to quartiles of plasma adiponectin are illustrated in Table 1. Patients with higher quartiles of adiponectin were older at the time of diagnosis (P = 0.03), had lower BMI (P < 0.001), and presented with higher grade tumors (P = 0.005).
Baseline characteristics of study participants according to quartiles of prediagnostic plasma adiponectin
. | Quartiles of plasma adiponectin . | |||
---|---|---|---|---|
Characteristic . | 1 (n = 196) . | 2 (n = 134) . | 3 (n = 151) . | 4 (n = 140) . |
Age at diagnosis, mean (SD), y | 68.8 (8.7) | 69.6 (7.7) | 70.8 (8.6) | 71.3 (7.8) |
BMI, mean (SD), kg/m2 | 27.3 (4.4) | 26.8 (4.5) | 25.4 (3.6) | 24.3 (3.6) |
Gender, % | ||||
Males | 46 | 46 | 50 | 34 |
Females | 54 | 54 | 50 | 66 |
Race, % | ||||
White | 94 | 96 | 94 | 99 |
Other | 6 | 4 | 6 | 1 |
Smoking status, % | ||||
Current or past | 59 | 51 | 58 | 57 |
Never | 41 | 49 | 42 | 43 |
Physical activity, mean (SD), METs | 20.3 (18.3) | 22.7 (22.8) | 24.4 (25.9) | 24.4 (25.6) |
Regular use of aspirin or NSAIDs, ≥2 tablets/wk, % | 51 | 37 | 50 | 49 |
Year of CRC diagnosis, mean (SD) | 2000 (4) | 2000 (5) | 1999 (4) | 1999 (5) |
Tumor location, % | ||||
Colon | 91 | 85 | 88 | 89 |
Rectum | 9 | 15 | 12 | 11 |
Stage of disease, % | ||||
I | 27 | 30 | 30 | 24 |
II | 24 | 21 | 21 | 25 |
III | 20 | 19 | 17 | 24 |
IV | 12 | 15 | 13 | 16 |
Other/unknown | 17 | 16 | 19 | 12 |
Grade of differentiation, % | ||||
Well | 6 | 15 | 17 | 6 |
Moderate | 62 | 47 | 51 | 60 |
Poor/undifferentiated | 13 | 16 | 11 | 18 |
Unknown | 19 | 22 | 22 | 16 |
. | Quartiles of plasma adiponectin . | |||
---|---|---|---|---|
Characteristic . | 1 (n = 196) . | 2 (n = 134) . | 3 (n = 151) . | 4 (n = 140) . |
Age at diagnosis, mean (SD), y | 68.8 (8.7) | 69.6 (7.7) | 70.8 (8.6) | 71.3 (7.8) |
BMI, mean (SD), kg/m2 | 27.3 (4.4) | 26.8 (4.5) | 25.4 (3.6) | 24.3 (3.6) |
Gender, % | ||||
Males | 46 | 46 | 50 | 34 |
Females | 54 | 54 | 50 | 66 |
Race, % | ||||
White | 94 | 96 | 94 | 99 |
Other | 6 | 4 | 6 | 1 |
Smoking status, % | ||||
Current or past | 59 | 51 | 58 | 57 |
Never | 41 | 49 | 42 | 43 |
Physical activity, mean (SD), METs | 20.3 (18.3) | 22.7 (22.8) | 24.4 (25.9) | 24.4 (25.6) |
Regular use of aspirin or NSAIDs, ≥2 tablets/wk, % | 51 | 37 | 50 | 49 |
Year of CRC diagnosis, mean (SD) | 2000 (4) | 2000 (5) | 1999 (4) | 1999 (5) |
Tumor location, % | ||||
Colon | 91 | 85 | 88 | 89 |
Rectum | 9 | 15 | 12 | 11 |
Stage of disease, % | ||||
I | 27 | 30 | 30 | 24 |
II | 24 | 21 | 21 | 25 |
III | 20 | 19 | 17 | 24 |
IV | 12 | 15 | 13 | 16 |
Other/unknown | 17 | 16 | 19 | 12 |
Grade of differentiation, % | ||||
Well | 6 | 15 | 17 | 6 |
Moderate | 62 | 47 | 51 | 60 |
Poor/undifferentiated | 13 | 16 | 11 | 18 |
Unknown | 19 | 22 | 22 | 16 |
Abbreviations: CRC, colorectal cancer; METs, metabolic equivalent task score hours per week.
Plasma adiponectin and mortality
We examined the association of prediagnostic plasma adiponectin with colorectal cancer–specific and overall mortality. Compared with participants with levels of plasma adiponectin in the lowest quartile, the multivariate HRs for colorectal cancer–specific mortality were 1.18 (95% CI, 0.75–1.86) for those in the second quartile, 1.24 (95% CI, 0.78–1.98) for those in the third quartile, and 1.89 (95% CI, 1.21–2.97) for those in the highest quartile (Ptrend = 0.01; Table 2). The corresponding multivariate HRs for overall mortality for participants with adiponectin levels in the second quartile of adiponectin were 0.94 (95% CI, 0.65–1.36), 1.08 (95% CI, 0.75–1.56) for the third quartile, and 1.66 (95% CI, 1.15–2.39) for the highest quartile (Ptrend = 0.009), when compared with those with adiponectin levels in the lowest quartile (Table 2).
Overall and colorectal cancer-specific mortality according to quartiles of prediagnostic plasma adiponectin
. | Quartiles of plasma adiponectina . | . | |||
---|---|---|---|---|---|
Variable . | 1 . | 2 . | 3 . | 4 . | Ptrendb . |
Overall mortality (n = 621) | |||||
No. of events/no. at risk | 75/196 | 54/134 | 63/151 | 77/140 | |
Age-adjusted HR | 1 | 0.98 | 1.06 | 1.39 | 0.07 |
95% CI | Referent | 0.69–1.40 | 0.75–1.48 | 1.01–1.92 | |
Multivariate-adjusted HR | 1 | 0.91 | 1.04 | 1.51 | 0.024 |
95% CIc | Referent | 0.63–1.31 | 0.73–1.48 | 1.08–2.12 | |
Multivariate-adjusted HR | 1 | 0.94 | 1.05 | 1.70 | 0.007 |
95% CId | Referent | 0.65–1.35 | 0.73–1.52 | 1.18–2.44 | |
Multivariate-adjusted HR | 1 | 0.94 | 1.09 | 1.68 | 0.007 |
95% CIe | Referent | 0.65–1.36 | 0.75–1.57 | 1.16–2.41 | |
Multivariate-adjusted HR | 1 | 0.94 | 1.08 | 1.66 | 0.009 |
95% CIf | Referent | 0.65–1.36 | 0.75–1.56 | 1.15–2.39 | |
CRC-specific mortality (n = 621) | |||||
No. of events/no. at risk | 49/196 | 38/134 | 44/151 | 50/140 | |
Age-adjusted HR | 1 | 1.12 | 1.15 | 1.54 | 0.025 |
95% CI | Referent | 0.73–1.72 | 0.76–1.74 | 1.03–2.29 | |
Multivariate-adjusted HR | 1 | 1.10 | 1.18 | 1.77 | 0.013 |
95% CIc | Referent | 0.71–1.71 | 0.76–1.83 | 1.16–2.68 | |
Multivariate-adjusted HR | 1 | 1.17 | 1.24 | 1.97 | 0.006 |
95% CId | Referent | 0.75–1.85 | 0.78–1.97 | 1.25–3.09 | |
Multivariate-adjusted HR | 1 | 1.19 | 1.24 | 1.92 | 0.008 |
95% CIe | Referent | 0.76–1.87 | 0.78–1.97 | 1.23–3.02 | |
Multivariate-adjusted HR | 1 | 1.18 | 1.24 | 1.89 | 0.01 |
95% CIf | Referent | 0.75–1.86 | 0.78–1.98 | 1.21–2.97 |
. | Quartiles of plasma adiponectina . | . | |||
---|---|---|---|---|---|
Variable . | 1 . | 2 . | 3 . | 4 . | Ptrendb . |
Overall mortality (n = 621) | |||||
No. of events/no. at risk | 75/196 | 54/134 | 63/151 | 77/140 | |
Age-adjusted HR | 1 | 0.98 | 1.06 | 1.39 | 0.07 |
95% CI | Referent | 0.69–1.40 | 0.75–1.48 | 1.01–1.92 | |
Multivariate-adjusted HR | 1 | 0.91 | 1.04 | 1.51 | 0.024 |
95% CIc | Referent | 0.63–1.31 | 0.73–1.48 | 1.08–2.12 | |
Multivariate-adjusted HR | 1 | 0.94 | 1.05 | 1.70 | 0.007 |
95% CId | Referent | 0.65–1.35 | 0.73–1.52 | 1.18–2.44 | |
Multivariate-adjusted HR | 1 | 0.94 | 1.09 | 1.68 | 0.007 |
95% CIe | Referent | 0.65–1.36 | 0.75–1.57 | 1.16–2.41 | |
Multivariate-adjusted HR | 1 | 0.94 | 1.08 | 1.66 | 0.009 |
95% CIf | Referent | 0.65–1.36 | 0.75–1.56 | 1.15–2.39 | |
CRC-specific mortality (n = 621) | |||||
No. of events/no. at risk | 49/196 | 38/134 | 44/151 | 50/140 | |
Age-adjusted HR | 1 | 1.12 | 1.15 | 1.54 | 0.025 |
95% CI | Referent | 0.73–1.72 | 0.76–1.74 | 1.03–2.29 | |
Multivariate-adjusted HR | 1 | 1.10 | 1.18 | 1.77 | 0.013 |
95% CIc | Referent | 0.71–1.71 | 0.76–1.83 | 1.16–2.68 | |
Multivariate-adjusted HR | 1 | 1.17 | 1.24 | 1.97 | 0.006 |
95% CId | Referent | 0.75–1.85 | 0.78–1.97 | 1.25–3.09 | |
Multivariate-adjusted HR | 1 | 1.19 | 1.24 | 1.92 | 0.008 |
95% CIe | Referent | 0.76–1.87 | 0.78–1.97 | 1.23–3.02 | |
Multivariate-adjusted HR | 1 | 1.18 | 1.24 | 1.89 | 0.01 |
95% CIf | Referent | 0.75–1.86 | 0.78–1.98 | 1.21–2.97 |
aQuartiles of plasma adiponectin were calculated separately within each cohort based on the distribution among controls based on previously collected data.
bTests for linear trend were conducted using the median values for each quartile of adiponectin.
cMultivariate models were stratified by age at diagnosis and adjusted for date of blood draw, sex, site of primary cancer (colon vs. rectum), stage at diagnosis, and histologic grade of cancer (well, moderate, poor/unknown).
dMultivariate models were stratified by age at diagnosis, and adjusted for date of blood draw, sex, site of primary cancer (colon vs. rectum), stage at diagnosis, histological grade of cancer (well, moderate, poor/unknown), body mass index, physical activity, current or past smoking (yes or no), folate, calcium, alcohol, servings of red meat as a main dish, history of colorectal cancer in parent or sibling, and regular use of aspirin or NSAIDs (≥ 2 tablets per week).
eMultivariate models additionally adjusted for plasma CRP, soluble TNFRSF1B and IL6, as well as the factors listed in d.
fMultivariate models additionally adjusted for plasma IGF1/IGFBP3, as well as the factors listed in e.
We subsequently conducted analyses according to selected subgroups to assess whether the association between adiponectin and mortality differed according to other predictors of survival defined by age, sex, BMI, stage, grade, and site of primary cancer (Table 3). The influence of plasma adiponectin on colorectal cancer–specific mortality seemed to be more pronounced in patients with metastatic disease (HR, 3.02; 95% CI, 1.50–6.08), comparing extreme quartiles of adiponectin (Pinteraction = 0.026). In addition, we observed consistent results among patients who were diagnosed less than 2 years after blood collection and those who were diagnosed 2 or more years after blood collection (Pinteraction = 0.66). Comparing extreme quartiles of adiponectin, participants who were diagnosed less than 2 years after blood collection had a multivariate HR for colorectal cancer–specific mortality of 3.40 (95% CI, 1.03–11.2; Ptrend = 0.036). The corresponding multivariate HR for those who were diagnosed 2 or more years after blood collection was 1.77 (95% CI, 1.11–2.84, Ptrend = 0.023).
Subgroup analyses of colorectal cancer-specific mortality according to quartiles of adiponectin
. | Quartiles of plasma adiponectin . | . | . | |||
---|---|---|---|---|---|---|
Subgroup . | 1 . | 2 . | 3 . | 4 . | Ptrend . | Pinteraction . |
Low agea (n = 307) | ||||||
No. of events/no. at risk | 27/111 | 18/71 | 14/72 | 22/53 | ||
MV HR | 1 | 1.32 | 1.32 | 2.24 | 0.002 | |
95% CI | Ref | 0.71–2.46 | 0.67–2.60 | 1.21–4.12 | ||
High ageb (n = 314) | 0.19 | |||||
No. of events/no. at risk | 22/85 | 20/63 | 30/79 | 28/87 | ||
MV HR | 1 | 1.01 | 1.14 | 1.45 | 0.55 | |
95% CI | Ref | 0.54–1.90 | 0.63–2.07 | 0.81–2.62 | ||
Male (n = 274) | ||||||
No. of events/no. at risk | 15/90 | 16/61 | 24/75 | 16/48 | ||
MV HR | 1 | 1.65 | 2.09 | 2.27 | 0.006 | |
95% CI | Ref | 0.80–3.38 | 1.08–4.06 | 1.10–4.71 | ||
Female (n = 347) | 0.54 | |||||
No. of events/no. at risk | 34/106 | 22/73 | 20/76 | 34/92 | ||
MV HR | 1 | 0.99 | 0.84 | 1.24 | 0.14 | |
95% CI | Ref | 0.57–1.72 | 0.47–1.48 | 0.74–2.07 | ||
Stage I/II/III tumors (n = 536) | ||||||
No. of events/ no. at risk | 29/173 | 21/114 | 26/131 | 28/118 | ||
MV HR | 1 | 1.04 | 1.18 | 1.44 | 0.19 | |
95% CI | Ref | 0.59–1.86 | 0.69–2.04 | 0.83–2.50 | ||
Stage IV tumors (n = 85) | 0.026 | |||||
No. of events/no. at risk | 20/23 | 17/20 | 18/20 | 22/22 | ||
MV HR | 1 | 1.29 | 1.13 | 3.02 | 0.003 | |
95% CI | Ref | 0.62–2.67 | 0.52–2.45 | 1.50–6.08 | ||
Normal BMIc(n = 285) | ||||||
No. of events/no. at risk | 19/66 | 13/50 | 28/79 | 34/90 | ||
MV HR | 1 | 0.92 | 1.51 | 1.84 | 0.05 | |
95% CI | Ref | 0.44–1.95 | 0.80–2.87 | 0.99–3.42 | ||
Overweight or obesed (n = 336) | 0.77 | |||||
No. of events/no. at risk | 30/130 | 25/84 | 16/72 | 16/50 | ||
MV HR | 1 | 1.39 | 0.92 | 1.86 | 0.09 | |
95% CI | Ref | 0.80–2.41 | 0.47–1.78 | 0.98–3.53 | ||
Well or moderate grade (n = 410) | ||||||
No. of events/ No. at risk | 24/132 | 21/83 | 23/102 | 25/93 | ||
MV HR | 1 | 1.03 | 1.33 | 1.83 | 0.07 | |
95% CI | Ref | 0.54–1.96 | 0.71–2.49 | 0.97–3.45 | ||
Poor or undifferentiated grade (n = 89) | 0.50 | |||||
No. of events/no. at risk | 12/26 | 8/22 | 10/16 | 13/25 | ||
MV HR | 1 | 1.22 | 2.89 | 1.39 | 0.43 | |
95% CI | Ref | 0.46–3.26 | 1.11–7.57 | 0.58–3.34 | ||
Colon (n = 549) | ||||||
No. of events/no. at risk | 44/178 | 32/114 | 41/133 | 46/124 | ||
MV HR | 1 | 1.22 | 1.35 | 1.99 | 0.006 | |
95% CI | Ref | 0.75–1.98 | 0.83–2.19 | 1.25–3.17 | ||
Rectum (n = 72) | 0.89 | |||||
No. of events/no. at risk | 5/18 | 6/20 | 3/18 | 4/16 | ||
MV HR | 1 | 0.76 | 0.51 | 0.97 | 0.85 | |
95% CI | Ref | 0.22–2.66 | 0.11–2.26 | 0.24–3.83 |
. | Quartiles of plasma adiponectin . | . | . | |||
---|---|---|---|---|---|---|
Subgroup . | 1 . | 2 . | 3 . | 4 . | Ptrend . | Pinteraction . |
Low agea (n = 307) | ||||||
No. of events/no. at risk | 27/111 | 18/71 | 14/72 | 22/53 | ||
MV HR | 1 | 1.32 | 1.32 | 2.24 | 0.002 | |
95% CI | Ref | 0.71–2.46 | 0.67–2.60 | 1.21–4.12 | ||
High ageb (n = 314) | 0.19 | |||||
No. of events/no. at risk | 22/85 | 20/63 | 30/79 | 28/87 | ||
MV HR | 1 | 1.01 | 1.14 | 1.45 | 0.55 | |
95% CI | Ref | 0.54–1.90 | 0.63–2.07 | 0.81–2.62 | ||
Male (n = 274) | ||||||
No. of events/no. at risk | 15/90 | 16/61 | 24/75 | 16/48 | ||
MV HR | 1 | 1.65 | 2.09 | 2.27 | 0.006 | |
95% CI | Ref | 0.80–3.38 | 1.08–4.06 | 1.10–4.71 | ||
Female (n = 347) | 0.54 | |||||
No. of events/no. at risk | 34/106 | 22/73 | 20/76 | 34/92 | ||
MV HR | 1 | 0.99 | 0.84 | 1.24 | 0.14 | |
95% CI | Ref | 0.57–1.72 | 0.47–1.48 | 0.74–2.07 | ||
Stage I/II/III tumors (n = 536) | ||||||
No. of events/ no. at risk | 29/173 | 21/114 | 26/131 | 28/118 | ||
MV HR | 1 | 1.04 | 1.18 | 1.44 | 0.19 | |
95% CI | Ref | 0.59–1.86 | 0.69–2.04 | 0.83–2.50 | ||
Stage IV tumors (n = 85) | 0.026 | |||||
No. of events/no. at risk | 20/23 | 17/20 | 18/20 | 22/22 | ||
MV HR | 1 | 1.29 | 1.13 | 3.02 | 0.003 | |
95% CI | Ref | 0.62–2.67 | 0.52–2.45 | 1.50–6.08 | ||
Normal BMIc(n = 285) | ||||||
No. of events/no. at risk | 19/66 | 13/50 | 28/79 | 34/90 | ||
MV HR | 1 | 0.92 | 1.51 | 1.84 | 0.05 | |
95% CI | Ref | 0.44–1.95 | 0.80–2.87 | 0.99–3.42 | ||
Overweight or obesed (n = 336) | 0.77 | |||||
No. of events/no. at risk | 30/130 | 25/84 | 16/72 | 16/50 | ||
MV HR | 1 | 1.39 | 0.92 | 1.86 | 0.09 | |
95% CI | Ref | 0.80–2.41 | 0.47–1.78 | 0.98–3.53 | ||
Well or moderate grade (n = 410) | ||||||
No. of events/ No. at risk | 24/132 | 21/83 | 23/102 | 25/93 | ||
MV HR | 1 | 1.03 | 1.33 | 1.83 | 0.07 | |
95% CI | Ref | 0.54–1.96 | 0.71–2.49 | 0.97–3.45 | ||
Poor or undifferentiated grade (n = 89) | 0.50 | |||||
No. of events/no. at risk | 12/26 | 8/22 | 10/16 | 13/25 | ||
MV HR | 1 | 1.22 | 2.89 | 1.39 | 0.43 | |
95% CI | Ref | 0.46–3.26 | 1.11–7.57 | 0.58–3.34 | ||
Colon (n = 549) | ||||||
No. of events/no. at risk | 44/178 | 32/114 | 41/133 | 46/124 | ||
MV HR | 1 | 1.22 | 1.35 | 1.99 | 0.006 | |
95% CI | Ref | 0.75–1.98 | 0.83–2.19 | 1.25–3.17 | ||
Rectum (n = 72) | 0.89 | |||||
No. of events/no. at risk | 5/18 | 6/20 | 3/18 | 4/16 | ||
MV HR | 1 | 0.76 | 0.51 | 0.97 | 0.85 | |
95% CI | Ref | 0.22–2.66 | 0.11–2.26 | 0.24–3.83 |
NOTE: Multivariate models were adjusted for age at diagnosis, date of blood draw, sex, site of primary cancer (colon vs. rectum), stage at diagnosis, histological grade of cancer (well, moderate, poor/unknown), BMI, physical activity, current or past smoking (yes or no), folate, calcium, alcohol, servings of red meat as a main dish, history of colorectal cancer in parent or sibling, regular use of aspirin or NSAIDs (≥2 tablets per week), CRP, soluble TNFRSF1B, IL6, and plasma IGF1/IGFBP3. For each stratified analysis, the stratification variable was omitted from the model.
aLow age defined as below the median at blood draw (<60 for women and <68 for men).
bHigh age defined as above the median at blood draw (≥60 for women and ≥68 for men).
cNormal BMI defined as <25.0 kg/m2 based on World Health Organization (WHO) classification.
dOverweight and obese defined as ≥25.0 kg/m2 based on WHO classification.
Discussion
In this prospective cohort study, we demonstrated that higher levels of prediagnostic circulating adiponectin were associated with adverse survival among patients with colorectal cancer, with an approximately 2-fold higher risk of colorectal cancer–specific mortality and 1.7-fold higher risk of overall mortality, after adjusting for other potential determinants of mortality, including BMI. These results suggest that plasma adiponectin is likely to be an independent prognostic factor. Further subgroup analyses revealed that participants with metastatic disease appeared to have the greatest increase in risk of colorectal cancer–specific mortality.
The potential pleiotropic roles of adiponectin have generated much controversy. Adiponectin has been demonstrated to possess anticarcinogenic effects via both direct and indirect mechanisms. In vitro studies have shown that activation of the AMPK pathway in cancer cells, with consequent downregulation of mTOR and increased expression of cyclin-dependent kinase inhibitors P21 and P27 is responsible for its antiproliferative and apoptotic effects (14). Furthermore, adiponectin exerts a direct inhibitory effect on the PI3K/AKT pathway, an important intracellular signaling pathway responsible for regulating cell cycling, proliferation, and survival (15). Its indirect actions comprise of modulation of insulin sensitization and inflammation. Chronic hyperinsulinemia and insulin resistance have been established as one of the etiologic links between obesity and colon cancer (16). Adiponectin has been demonstrated to exert a profound insulin-sensitizing effect through activation of AMPK and peroxisome proliferator-activated receptor alpha (PPARA) pathways in xenograft models, with resultant inhibition of tumor growth and angiogenesis (38). In addition, adiponectin mediates the production of anti-inflammatory cytokines such as interleukin 10 (IL10) and metalloproteinase-1 inhibitor, inhibition of proinflammatory chemokines and adipokines such as IL6 and tumor necrosis factor (TNF), inhibition of myelomonocytic precursor cells (mediators of innate immunity), and downregulation of T- and B-cell recruitment, all of which serve to impede inflammation-induced oncogenesis (17).
In contrast, other studies have reported the procarcinogenic properties of adiponectin. It has been shown to stimulate the production of proinflammatory cytokines such as CXCL8, colony stimulating factor 2 (granulocyte-macrophage) (CSF2), and inhibit apoptosis via activation of PRKAA2/SIRT1/PPARGC1A signaling pathway (19). In addition, it has been demonstrated to possess proangiogenic activity in mouse models (21). Finally, adiponectin has been shown to promote colonic cell proliferation in a dose-dependent manner (18).
Several epidemiologic studies have demonstrated that higher circulating adiponectin is associated with a significant reduction in the risk of developing colorectal cancer (22–25). Results from a meta-analysis of 9 studies demonstrated a significant inverse relationship between plasma adiponectin and colorectal cancer (OR, 0.91, 95% CI, 0.83–1.00; P = 0.04; ref. 24). This is consistent with the results from another meta-analysis of 13 studies that included 2,632 cases of colorectal cancer or adenoma and 2753 healthy controls (23). However, a recent meta-analysis of 26 studies demonstrated that an increment of 1 mg/L in circulating adiponectin was significantly associated with a 20% to 30% increased risk of colorectal cancer (39).
Studies assessing the potential impact of circulating adiponectin on colorectal cancer survival are comparatively limited. The only prospective study revealed that plasma adiponectin was not a significant predictor of colorectal cancer survival (HR, 1.00; 95% CI, 0.97–1.04; P = 0.94; ref. 26). However, in this study, plasma adiponectin was measured at the time of diagnosis of colorectal cancer rather than prediagnostically. To our knowledge, our study is the first to examine the relationship between plasma adiponectin measured prior to colorectal cancer diagnosis and survival in patients with colorectal cancer.
Because prior studies have suggested a protective effect of higher circulating adiponectin on the development of incident colorectal cancer, our results showing that prediagnostic circulating adiponectin is significantly associated with worse prognosis among patients with colorectal cancer may initially appear inconsistent. However, similar findings have been reported for other gastrointestinal cancers such as HCC. Recently, Siegal and colleagues demonstrated that higher serum adiponectin level at the time of diagnosis of HCC is associated with worse OS (HR, 1.90; 95% CI, 1.05–3.45; P = 0.03; ref. 27).
Several hypotheses may account for our findings. First, it is possible that a high level of circulating adiponectin is associated with a lower incidence of colorectal cancer, but individuals who develop colorectal cancer despite elevated adiponectin levels may be predisposed to tumors that arise through distinct mechanisms. These tumors may have adverse features associated with poor prognosis that could explain our findings. Second, although several in vitro and in vivo studies suggest that adiponectin exerts an anti-inflammatory effect (40, 41), additional studies have refuted this notion, particularly in the setting of chronic conditions such as inflammatory bowel disease (42) and rheumatoid arthritis (43). Hence, it is possible that the inflammatory microenvironment generated by the predominance of proinflammatory factors, which is mediated by adiponectin, in the setting of individuals with cancer may facilitate tumor proliferation and metastasis, thus leading to worse survival (44, 45). Furthermore, a study conducted in adiponectin transgenic mice showed that elevated levels of adiponectin do not protect against colorectal cancer development (46). Third, emerging evidence has demonstrated that adiponectin exerts potent antiapoptotic effect via activation of the PRKAA2/SIRT1/PPARGC1A pathway, leading to increased mitochondrial gene expression and cell proliferation (19). Finally, adiponectin has been associated with a lower risk of development of fatty liver (47). Several studies have supported a potential protective effect conferred by fatty liver against hepatic metastasis via inhibition of angiogenesis (48). Hence, patients with low adiponectin levels may be relatively protected against the development of hepatic metastasis and thus better prognosis in the absence of metastatic disease elsewhere.
The strengths of our study include large sample size, prospective design, detailed epidemiologic data, high follow-up rates, and high accuracy of self-reported data, as our cohort participants compose of health professionals. Several limitations warrant comment in our study. We were unable to distinguish between HMW and LMW adiponectin. There is some suggestion that HMW adiponectin, being the more biologically active form, may be a better predictor of colorectal cancer risk (49). The two isoforms also possess inherent differences with respect to inflammation, with HMW adiponectin being proinflammatory and LMW adiponectin anti-inflammatory. Furthermore, information on cancer recurrence and relapse were not available in these cohorts. However, colorectal cancer–specific survival should be a reasonable surrogate for recurrence as the median survival for recurrent colorectal cancer was approximately 10 to 12 months during the time period of this study. We also had limited data on treatment. However, it is unlikely that the heterogeneity in treatment during the time period under study could explain the observed results. The treatment paradigm did not change significantly during this period and receipt of chemotherapy was largely defined according to stage, which we accounted for in our analysis. In addition, our analysis used a single measurement of adiponectin drawn prior to colorectal cancer diagnosis. However, previous studies have established that adiponectin levels remain stable over time (36), and given the prospective design of our study, any measurement error in adiponectin level would likely attenuate our observed associations. We also did not measure adiponectin levels after diagnosis. Thus, we were unable to examine the association of adiponectin with survival according to levels prediagnosis compared with postdiagnosis. Finally, the generalizability of our results to other populations may be limited as our participants were restricted to healthcare professionals.
In conclusion, our study provides evidence that higher levels of prediagnostic plasma adiponectin are associated with increased risk of colorectal cancer–specific and overall mortality in patients with colorectal cancer. Our findings support the notion that plasma adiponectin may play a distinct biological role in the progression of colorectal cancer compared with its development. Additional prospective studies are warranted to fully elucidate the underlying biologic mechanisms of adiponectin and its associated pathways and determine its clinical and prognostic use in colorectal cancer.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: D.Q. Chong, C.S. Fuchs, A.T. Chan
Development of methodology: D.Q. Chong, R.S. Mehta, C.S. Fuchs, E.L. Giovannucci, A.T. Chan
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Song, C.S. Fuchs, A.T. Chan
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D.Q. Chong, R.S. Mehta, M. Song, D. Kedrin, J.A. Meyerhardt, K. Ng, K. Wu, C.S. Fuchs, E.L. Giovannucci, S. Ogino, A.T. Chan
Writing, review, and/or revision of the manuscript: D.Q. Chong, M. Song, D. Kedrin, J.A. Meyerhardt, K. Ng, K. Wu, C.S. Fuchs, E.L. Giovannucci, S. Ogino, A.T. Chan
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.T. Chan
Study supervision: S. Ogino, A.T. Chan
Acknowledgments
The authors thank the participants and staff of the NHS and the HPFS, for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY.
Grant Support
This work was supported by U.S. NIH grants [P01 CA 087969 and R01 CA49449 (to S.E. Hankinson), UM1 CA167552 and P01 CA55075 (to W.C. Willett), P50 CA127003 (to C.S. Fuchs), R01 CA151993 and R35 CA197735 (to S. Ogino), K24 DK 098311 and R01 CA137178 (to A.T. Chan)]. D.Q. Chong is a recipient of the Singhealth Health Manpower Development Plan (HMDP) fellowship award from Singapore. R.S. Mehta is a Howard Hughes Medical Institute Medical Research Fellow and an AGA–Eli and Edythe Broad Student Research Fellow.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.