Assessment of Adipokine and Inflammation Biomarkers with Cancer Risk among Chinese Men: A Prospective Cohort Study.

Introduction
Obesity and obesity-related diseases, including several cancers, have become a global epidemiologic concern.1 The global prevalence of the overweight and obese men is nearly double in 2021 (43.4%) compared to 1990 (27.6%).2 While the obesity rates in China are much lower compared to Western countries,2 they also have gone substantial increase,3 with the prevalence of adult obesity going from 3.1% (2004) to 16.4% (2023),4 and overweight-obesity rates in men projected to increase by 70% in 2030.5 Obesity has been linked to 13 types of cancer, including stomach, colorectal, gallbladder, kidney, liver, meningioma, multiple myeloma, esophageal, ovary, pancreatic, endometrium, thyroid, and breast.6
Lung, liver, stomach, and colorectal cancer (CRC), are the top four most common cancers and causes of cancer mortality in China.7 The age standardized incidence rates for stomach (29.5/100,000), liver (27.6/100,000), and lung cancer (47.8/100,000) in 2020 were much higher for Chinese men, compared to U.S. men (5.3, 10.4, 36.3/100,000 for stomach, liver and lung cancer, respectively); though CRC was similar in Chinese (28.6/100,000) and U.S. (28.7/100,000) men.8 In 2015, it was estimated that 416,369 lung; 179,481 stomach; 175,764 CRC; and 150,965 liver cancer cases were diagnosed among men in China.9 The number of these cancer cases was projected to increase at a steady rate, so by the year 2030 there would be 603,425 lung; 324,871 stomach; 232,974 CRC; and 224,613 liver cancer cases.9
Obesity, smoking, and unhealthy diet are the major contributors for these 4 major cancers.7 We have previously shown that obesity in men was associated with a 74% increased risk of CRC in Urban Shanghai.10 A systematic review and meta-analysis by Sohn et al. found that overweight-obese adults showed a 36% to 3-fold increased risk of liver cancer, respectively.11 In a pooled analysis of more than 1.6 million people from cohort the United States, Europe, and Asia, we found central obesity was associated with increased lung cancer risk while overweight or obese was inversely associated with the risk.12 Additionally, a systematic review and meta-analysis of over 14 million people found a 16% increased risk of stomach cancer among the overweight and a 22% increased risk among obese people.13
Several mechanisms link obesity with cancer development, such as altered gut microbiota, adipokine imbalance, chronic inflammation, coagulation activation, sex hormone dysregulation, insulin resistance and hyperinsulinemia.6 These alterations, individually or jointly, lead to increases in cell proliferation, angiogenesis, and reductions in apoptosis, subsequently increasing the individual’s risk of developing cancer.6,14,15 Obesity has been linked to elevated levels of C-reactive protein (CRP),16,17 leptin, resistin, interleukin-6 (IL-6), plasminogen activator inhibitor-1 (PAI-1), monocyte chemoattractant protein-1 (MCP-1), tumor necrosis factor-alpha (TNFα), and lower levels of adiponectin.18 Systemic inflammatory biomarkers may serve as possible predictors for cancer risk and prognosis.19,20 A population-based prospective cohort study found that inflammatory biomarker scores, namely CRP in combination with elevated leukocyte counts, was associated with increased cancer risk and mortality.21 A review conducted by Garcia-Anguita et al. found that elevated CRP levels were associated with a 27%−65% increased risk of CRC22 among various cohorts from the United States,23–25 Finland,26 Europe,27 Korea,28 and China.29 Elevated levels of leptin30 and insulin31 were also associated with increased CRC risk. Additionally, prior epidemiologic studies found that TNFα and cytokine interleukin-1 beta (IL-1β) were associated with increased risk of lung cancer.32
However, the associations of these obesity and inflammation biomarkers with cancer risk have not been well investigated among Chinese men who have a relative low prevalence of obesity but a high prevalence of smoking. For example, in the Shanghai Men’s Health Study (SMHS), a large population-based cohort (n=61,469),33 the prevalence of obesity was 2.5% and the current smoking rate was 58.7%.34 According to the International Agency for Research on Cancer (IARC), China is one of three countries that account for over 50% of the world’s male smokers.35 Smoking, although inversely associated with overall obesity,36 is associated with central obesity and high levels of inflammation markers, such as IL-6, interleukin-8 (IL-8), IL-1ꞵ, TNFα,37 and CRP.38 To address this research gap, we comprehensively investigated the associations of multiple adipokines and inflammation biomarkers with the risk of total cancers in addition to lung, stomach, CRC, and liver cancers (the top cancers) among a cohort of SMHS participants.

Materials and Methods

Study Population and Design
Included in the study are 4,051 (6.6%) participants of the SMHS who donated pre-diagnosis fasting plasma samples at baseline. The SMHS is a population-based cohort study which enrolled 61,469 men aged 40–74 during 2002–2006 from 8 communities in Urban Shanghai and were followed for cancer incidence and mortality, with a median follow-up time of 12.17 years. Men with pre-existing cancers at baseline or non-permanent residents of Shanghai were not eligible for participation in the SMHS. The response rate at baseline recruitment was 74% with above 90% responses at each follow-up. More information regarding the SMHS design and participants’ characteristics can be found elsewhere.33

Sample Collection & Assays
At study enrollment, we collected a 10-ml blood sample from each of the 46,105 consenting study participants (75.1%). All samples were kept at 4°C, processed within 6 hours, and stored at −80°C. Plasma samples were used for biomarker analyses. At biospecimen collection, information was collected on date and time of sample collection, time of last meal, intake of selected foods, smoking, and use of medications over the past 24 hours and during the preceding week.33 Among subjects who donated a blood sample at baseline, 4,051 subjects who donated a fasting blood sample as well as had no diabetes history and a negative urine glucose test, a caloric intake per day that was not ≤800 kcal or ≥4,200 kcal, and had lipid measurements were selected for the current study (Figure 1). Plasma levels of leptin, resistin, adiponectin, plasminogen activator inhibitor type-1 (PAI-1), interleukin-6 (IL-6), interleukin-8 (IL-8), interleukin-1 beta (IL-1β), tumor necrosis factor-alpha (TNFα), monocyte chemoattractant protein-1 (MCP-1), and insulin were measured by Luminex assays, and C-reactive protein (CRP) was measured by ACE Clinical Chemistry Systems following the manufacturer’s guidelines. Commercial multiplex Luminex assays were applied in our study because of their ability to measure multiple analytes simultaneously with high sensitivity and precision and require a small sample volume. The coefficients of variation from biomarkers under study were leptin (9.1%), resistin (15.8%), adiponectin (8.2%), PAI-1 (8.8%), IL-6 (16.3%), IL-8 (12.5%), IL-1β (17.8%), TNFα (11.0%), MCP-1 (8.3%), insulin (9.0%), and CRP (6%−8%). The missing rates (due to assay failure or no samples) for biomarkers under study are low (missing range n = 1–52, 0.02%−1%), with the exception of IL-1β (n = 212 missing, 5%), and subjects with missing data were excluded from the analysis.
New cancer cases, vital status, and cause of death were identified through annual linkages with the Shanghai Cancer Registry and Shanghai Vital Status Registry and verified in-house by the study research team. The cancers under study were classified in accordance with the International Classification of Disease, 9th Revision (ICD-9) criteria: all cancers (140–208), lung cancer (162), stomach cancer (151), colorectal cancer (153, 154), and liver cancer (155). Due to the small sample size, esophageal cancer could not be evaluated individually.

Covariates
Demographic characteristics, lifestyle habits, and health status were ascertained by in-person interviews with structured questionnaires. Covariates adjusted for in this study included age at enrollment, income in yuan/capital/month (0 = <500, 1 = 500–999, 2 = 1000–1999, 3 = 2000+), education (0 = no formal education, elementary school, and junior high school; 1 = high school, 2 = professional high education and college or above), Chinese Healthy Dietary Index score (derived based on dietary habits during the past 12-months, with sum scores ranging from 0 “lowest” to 45 “highest” adherence),39 hepatitis (0 = no, 1 = yes), smoking (0 = never, 1 = former, 2 = current), pack-years smoking (packs/day * years of smoking), body mass index (BMI, kg/m2), regular physical activity (once/week at least for 3 months continuously over the past 5 years, 0 = no, 1 = yes), hypertension (0 = no, 1 = yes), and using Nonsteroidal Anti-Inflammatory Drugs (NSAIDs, 0 = no, 1 = yes).

Ethical Approval
The study was reviewed and approved by the Institutional Review Boards of Vanderbilt University Medical Center and the Shanghai Cancer Institute. The study was conducted in accordance with the ethical guideline of the International Ethical Guidelines for Biomedical Research Involving Human Subjects, and written informed consent was obtained from all study participants.

Statistical Analysis
Cancer cases were censored at date of diagnosis and non-cancer participants were censored at date of death or end of follow-up date (December 31, 2016), whichever occurred first. Descriptive statistics were used to describe baseline demographics and characteristics of the study population. Shapiro-Wilk’s test for normality and histograms were used to examine the distribution of biomarkers. Biomarkers with concentrations below detectable levels were assigned scores by half of the lowest observed value. Log transformations were applied for all biomarkers to improve normal distribution. Analyses of biomarker levels were assessed categorically via tertile distribution and continuously after log transformation. Spearman’s Rho was used to evaluate correlations between biomarkers under study. Considering competing risks, Fine and Gray’s sub-distribution multivariable models were applied to estimate adjusted hazard ratios (aHR) and 95% confidence intervals (CI) for cancer in association with biomarkers, with adjustment for age at enrollment, education, income, Chinese Healthy Dietary Index, pack-years of smoking, BMI, physical activity, hypertension, and NSAID use. Hepatitis was further adjusted for in the liver cancer models. Multivariable Cox regression models were used for analyses involving any diagnosed cancer. Analyses stratified by time interval between blood draw and cancer diagnosis (<5 vs ≥5 years) were conducted to evaluate whether the biomarker-cancer risk varied by time window since blood draw. All statistical tests were based on two-tailed probability and a significance level set at alpha (α) <0.05. All analyses were performed using SAS office analysis (Version 7.12).

Data Availability:
The data generated in this study are available upon request from the Corresponding Author.

Results
Described in Table 1 are the characteristics of the study population (n=4051). The mean age of men at enrollment was 51.0 (range 40.0–74.9). The majority of men (41.9%, n=1,698) had an income of 500–999 yuan/capital/month and a high school education (50.0%, n=2027). 75.5% of participants were current smokers (n=3060) and the average pack-years of smoking among current and former smokers was 25.2 (range 0.1–187.4 pack-years). The average Chinese Healthy Dietary Index score was 32.4 (range 9.0–43.5) and BMI was 23.3 kg/m2 (range 13.9–39.5 kg/m2). 78.3% of participants (n=3171) were not regular exercisers (once/week at least, for 3 months continuously), 79.3% had no hypertension (n=3,212), 70.3% of participants were not NSAIDs users (n=2,848), and 2.5% had a history of hepatitis (n=100). During 15-years of follow-up (median follow-up time of 12.77 years), 319 cancer cases, including 85 lung, 41 stomach, 17 liver, and 61 CRC cases occurred.
Table 2 shows the correlation matrix for all biomarkers under study. Most biomarkers were significantly correlated with each other with the correlation coefficient ranging from −0.24 to 0.54. The highly positively correlated biomarkers include leptin-insulin (ρ = 0.54), TNFα-IL-8 (ρ =0 .40), MCP-1-IL-8 and resistin-PAI-1 (ρ = 0.38 for both), TNFα-MCP-1 and TNFα-IL-6 (ρ = 0.33 for both), and leptin-PAI-1 (ρ =0.30). The biomarkers highly inversely correlated include adiponectin-leptin and adiponectin-insulin (ρ = −0.24 for both) IL-1β- resistin and PAI-1 (ρ = −0.22).
Presented in Table 3 are results from multivariable analyses on biomarkers and cancers under study. After adjustment for age at enrollment, education, income, Chinese Healthy Dietary Index, pack-years smoked, BMI, physical activity, hypertension, and use of NSAIDs (and hepatitis for liver cancer models), when comparing the 3rd vs 1st tertiles, higher levels of TNFα were associated with increased risk of lung cancer (aHR=1.78, 95%CI=1.04–3.05, Ptrend=0.03). Higher levels of leptin (aHR=3.00, 95%CI=1.23–7.25, Ptrend=0.01), insulin (aHR=2.26, 95%CI=1.04–4.92, Ptrend <0.05), and CRP (aHR=2.24, 95%CI=1.16–4.32, Ptrend=0.01) were associated with increased risk of CRC, and higher levels of insulin were associated with increased risk of liver cancer (aHR=3.87, 95%CI=1.07–14.0, Ptrend=0.04). Additionally, per-log-unit increase of TNFα was associated with increased risk of total cancer (aHR=1.23, 95%CI=1.00–1.52) and lung cancer (aHR=1.45, 95%CI=1.00–2.15). Per-log-unit increase of insulin was associated with increased risk of CRC (aHR=1.46, 95%CI=1.04–2.05) and liver cancer (aHR=1.86, 95% CI=1.03–3.38). Per-log-unit increase of CRP was associated with increased risk of CRC (aHR=1.20, 95%CI=1.00–1.44), and per-log-unit increase in IL-8 was associated with liver cancer (aHR=1.81, 95%CI=1.00–3.30). Although adiponectin was not statistically associated with liver cancer when comparing the 3rd vs 1st tertile, per-log-unit increase in adiponectin showed a significant association with liver cancer risk (aHR=3.27, 95%CI=1.64–6.52). No significant associations were observed between MCP-1, resistin, PAI-1, IL-6, and IL-1β and cancers under study (Table 3).
Presented in Table 4 are results of analyses for cancers developed within 5 years following the blood draw. We found that higher IL-6 levels were associated with increased total cancer incidence (aHR=1.92, 95%CI=1.08–3.40 for 3rd vs 1st tertile, Ptrend=0.03) while higher levels of TNFα (aHR=3.71, 95%CI=1.05–13.10, Ptrend=0.03), IL-8 (aHR=3.72, 95%CI=1.06–13.1, Ptrend=0.02), and IL-1β (aHR=2.85, 95%CI=1.00–8.10, Ptrend=0.04) were associated with increased lung cancer risk. IL-1β was also associated with increased stomach cancer risk when comparing the 3rd vs 1st tertile (aHR=5.85, 95%CI=1.50–22.8, Ptrend=0.01). Additionally, per-log-unit increase in TNFα was associated with increased risk of total cancer (aHR=1.67, 95%CI=1.14–2.46), lung (aHR=1.98, 95%CI=1.01–3.86), and liver (aHR=2.68, 95%CI=1.11–6.47) cancers. Furthermore, per-log-unit increase of IL-8 level was associated with increased liver cancer (aHR=2.38, 95%CI=1.08–5.08) risk, while IL-1β was associated with stomach cancer (aHR=1.46, 95%CI=1.08–1.97). PAI-1 was the only biomarker to show an inverse association with cancer risk (stomach cancer, aHR=0.66, 95%CI=0.46–0.94). No significant associations were observed for leptin, insulin, resistin, adiponectin, MCP-1, and CRP for cancer risk within 5 years following blood draw (Table 4).
The association between inflammation biomarkers and cancer risk ≥5 years after blood draw are presented in Table 5. Higher levels of leptin were significantly associated with an increased risk of CRC (aHR=3.39, 95%CI=1.26–9.10, Ptrend=0.01 for 3rd vs 1st tertile). Continuous variable analyses showed that per-log-unit increase of adiponectin was associated with an increased risk of liver cancer (aHR=4.78, 95%CI=1.87–12.3), and per-log-unit increase in insulin was associated with an increased risk of CRC (aHR=1.57, 95%CI=1.08–2.27). Per-log-unit increase in PAI-1 (aHR=0.71, 95%CI=0.52–0.97) and insulin (aHR=0.45, 95%CI=0.21–0.96) were the only biomarkers significantly associated with a decreased risk of cancer (lung cancer). No significant associations were observed for TNFα, resistin, IL-8, IL-6, MCP-1, IL-1β, or CRP for cancer risk ≥5 years after blood draw (Table 5).

Discussion
In a sub-cohort from the SMHS, a population-based cohort study, we prospectively investigated the associations of circulating levels of adipokines and inflammation biomarkers with total cancer risk, as well as risk of lung, liver, colorectal, and stomach cancers.

Colorectal Cancer
CRP is a well-known inflammatory marker and contributes to tumorigenesis in higher concentrations.40 Chronic elevated CRP can impact the immune cell functionality in addition to obstructing the metabolic process that contributes to disorders such as obesity and insulin resistance.40 The compounding effect in metabolic changes and elevated CRP can disrupt the gut microbiota which can contribute to tumor-cell growth, consequently increasing risk of CRC.40 Our findings that higher levels of CRP were associated with over a 2-fold (3rd vs 1st tertile) increased risk of CRC, and per-log-unit increase of CRP was associated with a 20% higher risk, are supported by prior epidemiologic studies, as summarized in the Garcia-Anguita et al. systematic review22 and a recent meta-analysis.41 A positive association between CRP and CRC were seen in several cohort studies from the United States,23–25 Finland,26 Europe,27 Korea,28 and China,29 although null associations were also reported,42,43–45 particularly among women.46,47 Il’yasova et al. observed a 64% CRC risk increment for each unit of CRP increase.25 These findings, including those from our study, suggest that CRP may serve as a potential risk marker for identifying those at high risk of developing CRC for close surveillance.
Hyperinsulinemia has been shown to promote cell proliferation towards the progression of colon cancer.48 One potential pathway of hyperinsulinemia involves insulin resistance, which is the inability or rapid decline of cells to respond to insulin in the body, which in turn can impact the metabolic process and increase the risk of obesity and developing of CRC.49 The hyperinsulinemia and insulin resistance mechanisms are interconnected and can be accelerated through inflammation.50 In our study, we found that increased insulin levels correlated with inflammatory factors, such as TNFα, CRP, and IL-1β. Additionally, we found that per-log-unit increase of insulin was associated with a 46%−57% increase in CRC risk. This corresponds with Xu et al.51 observing a 42% risk of CRC and Limburg et al.’s52 findings linking high serum insulin levels to CRC risk in male smokers,52 as well as Zhang et al.’s study31 and a meta-analysis.53
Leptin plays an important role in energy metabolism, regulates appetite, and is a known risk factor for obesity and CRC, as well as an inflammation inducer.54 Leptin is secreted by the adipose tissue;55 leptin induces proliferation and invasiveness in CRC cell lines through activations in MAPKs, PI3K, NF-κB, and STAT3, and is known to be associated with the secretion of IL-6 and TNFα.55 In our study, leptin levels were significantly correlated with higher levels of IL-6 and TNFα. We found elevated leptin levels were associated with a 3-fold increased CRC risk in men, consistent with findings from previous studies.55,56 However, a meta-analysis of multiple case-control studies showed null results for leptin and CRC associations.30 It is worth pointing out that our study our study applied a prospective study design and leptin was measured in pre-diagnostic plasma. In addition, we found that the leptin-CRC association was only observed during the window of time 5-years before cancer diagnosis or earlier. Thus, leptin is more likely a risk biomarker and would be difficult to discover in retrospective case-control studies.

Lung Cancer
TNFα is a well-known inflammation marker that is increased among smokers. TNFα is associated with tumor cell survival, proliferation, and metastasis.57 Elevated TNFα levels were observed among lung cancer patients with higher cancer stages.58 We found that per-log-unit increase in TNFα level was associated with a 45% increased risk of lung cancer and a doubled risk (98% risk per-log-unit) when analyses was restricted for lung cancer patients diagnosed <5 years after blood draw. This closely corresponds with Il’yasova et al.’s study findings, in which the per-unit increase of TNFα levels was associated with a 67% (aHR = 1.67, 95% CI = 0.79–3.55) increase in lung cancer risk, although it was not statistically significant.25 Of note, Il’yasova et al.’s study did not adjust for smoking, diet, BMI, hypertension, or use of NSAIDs. Peng et al.59 observed statistically significant correlations between elevated TNFα levels and occurrences of lung cancer (r=0.40). However, a null association was reported in two other studies.60,61 More studies are needed to further investigate the association between TNFα and lung cancer risk.
Interleukins, such as IL-6, IL-8, and IL-1β are other proinflammatory markers that are present among smokers and play an important role in lung cancer risk.62 IL-6 enhances angiogenesis, tumor invasion, growth, and metastasis,57 while IL-8 stimulates angiogenesis, cancer cell motility, and dissemination by stimulating their migration and invasion through the extracellular matrix.63 IL-1β encourages angiogenesis, immune evasion, and tumor growth and induces inflammation and facilitate tumor formation.57 In our study, we found that higher IL-6, IL-8, and IL-1β levels were associated with an elevated but not statistically significant risk of lung cancer. However, once analyses were restricted to within 5-years after blood draw, IL-8 and IL-1β were associated with a 3.7-fold and 2.9-fold increased risk of lung cancer when comparing the highest vs lowest tertiles, respectively. Elevated IL-6 level was associated with lung cancer risk in minimally adjusted models but not in fully adjusted models. The IL-8 findings correspond with Pine et al.’s study that found higher IL-8 levels were associated with a risk of lung cancer diagnoses within 2 and 5 years before diagnosis64 and impacted overall survival.63 Additionally, our study partially agrees with Park et al.’s study that observed higher IL-1β levels associated with a 61% increased risk of lung cancer.60,61 Future studies should consider evaluating whether TNFα and interleukins (IL-6, IL-8, and IL-1β) influence the progression of lung cancer as elevated TNFα, IL-6, IL-8, and IL-1β levels were associated with the invasiveness of lung adenocarcinoma.62

Liver Cancer
Adiponectin plays a complex role in the metabolic process of liver cancer.65 A meta-analysis by Zhang et al. found several studies that identified that elevated adiponectin was associated with liver cancer, compared to healthy control groups.66 Zhang et al. also reported a significant linear relationship between per-unit increase in adiponectin levels and liver cancer risk, as-well-as decreased overall survival.66 In our study, we found that per-log-unit increase in adiponectin levels was associated with a 3.3-fold increased risk of liver cancer, which increased to nearly a 5-fold risk when analysis was done among cases diagnosed >5-years after blood draw. Our study findings correspond with a Mendelian randomization study from Jiang et al. which found that a 10% incremental increase in genetically determined levels of adiponectin was associated with a 26% increased risk (aHR=1.26, 95%CI=1.09–1.44) of hepatocellular carcinoma.67
IL-8 is capable of promoting liver cancer cell invasion and metastasis through CXC chemokine receptors 1 and 2.68 In our study, we found that per-log-unit increase in IL-8 levels was associated with a 1.8-fold increased risk of liver cancer, which increased to 2.4-fold when analysis was done among cased diagnosed within 5 years after blood draw. These findings correspond with a meta-analysis by Shakiba et al. that found serum IL-8 levels were higher among patients with hepatocellular carcinoma compared to the healthy control group.69
Liver cancer is linked with one potential mechanism that involves the elevated levels of circulating insulin or insulin-like growth factors.70,71 In our study, we found that per-log-unit increase in insulin was associated with a 1.9-fold increased liver cancer risk, with the highest tertile level being associated with a near 4-fold increased risk. Our findings correspond with a study by Loftfield et al. that found elevated insulin levels were associated with a 3 to 7-fold increased risk of liver cancer.72 Our findings also correspond to Yin et al.’s study which found that higher insulin levels (comparing the highest vs lowest quartile) were associated with a nearly 2.5-fold increased risk of liver cancer.73

Stomach Cancer
Circulating IL-1β may promote the development and progression of stomach cancer by affecting growth, angiogenesis, proliferation, metastasis, and invasion of gastric cancer cells,74 as well as inhibiting gastric acid secretion and inducing epigenetic alterations.75 In our study, elevated IL-1β levels were associated with an elevated risk of stomach cancer when analysis was restricted to <5-years since blood draw (a 6-fold risk comparing 3rd vs 1st tertile, and 46% elevated risk per-log-unit increase of IL-1β). These findings are supported by evidence from genetic epidemiologic studies. In a meta-analysis assessing two polymorphisms of IL-1β (−511C/T and +3954C/T), Xu et al. found that IL-1β −511C/T was associated with a 57% risk of non-cardia stomach cancer (OR = 1.57, 95% CI: 1.06–2.31) and a 24% risk of intestinal stomach cancer (OR = 1.24, 95% CI: 1.04–1.49).76 He et al.77 also reported significant associations between polymorphisms in IL-1β and stomach cancer risk among a Chinese population.77
Our study has several strengths, including prospectively investigating multiple inflammation biomarkers in a large population-based cohort. Our study collected detailed information on lifestyle factors, disease history, Chinese Healthy Dietary Index, and other covariates which allowed for controlling for multiple potential confounders. Additionally, all blood samples included in the study were collected after at least 8 hours of fasting, minimizing the influence of dietary intake on these biomarkers. However, there are several limitations to be noted. First, blood samples from our study were collected at baseline. It is possible that regression dilution bias78 may be present due to potential changes of biomarker levels over time, leading to an underestimation of the biomarker-cancer associations towards the null.78 Our finding of some stronger biomarker-cancer associations during the 5-year period following blood draw, compared to those during the period beyond 5 years, may be due to a more accurate exposure assessment. However, as discussed above, these stronger associations could be due to the fact that these biomarkers, particularly inflammatory biomarkers, are results of cancer progression. Second, our study was conducted among men living in urban Shanghai. Thus, results may not be generalizable to other populations. Third, it should be noted that some cancers (such as liver) had small sample sizes, therefore affecting the precision of risk estimation. Last, it is possible that residual confounding, such as genetic factors and stress, could influence levels of biomarkers and cancer risk that could not be accounted for and may be present in this study.
In conclusion, this prospective cohort study found that high plasma levels of leptin, insulin, and CRP were associated with the risk of developing CRC; high adiponectin, insulin, and IL-8 levels were associated with liver cancer risk, and high TNFα levels with lung cancer risk. Our study suggests that incorporating CRP, TNFα, and possibly IL-6 and IL-1β in risk assessment may identify individuals at a high risk of developing cancer who may benefit from targeted cancer screening or close surveillance.