Risk of Cancer Mortality according to the Metabolic Health Status and Degree of Obesity

act


Introduction
Increasing rates of obesity and metabolic disturbance are becoming major public health problems in Korea (Wang et al., 2008;Finkelstein et al., 2012;Kang et al., 2014;Lee et al., 2014).Obesity is associated with numerous non-communicable diseases, such as type 2 diabetes mellitus, metabolic syndrome, cardiovascular diseases, and cancer.Notably, cancer is the leading cause of death and is associated with the largest disease burden in Korea (Lee et al., 2013;Jung et al., 2014).Many studies have suggested that obesity is related to all-cause mortality and cancer mortality (Calle et al., 2013;Flegal et al., 2013;Ozbek et al., 2013).Indeed, the risk of death from cancer increases with body mass index (BMI) (Jee et al., 2006;Lee et al., 2013).A previous report found that BMI was an independent predictor of the incidence and mortality of a variety of cancers including colon cancer (Moore et al., 2004;Renehan et al., 2008;Vucenik et al., 2012;Calle et al., 2013;Morrison et al., 2013).In addition, obesity-

Risk of Cancer Mortality according to the Metabolic Health Status and Degree of Obesity
Chang-Mo Oh, Jae Kwan Jun, Mina Suh* has examined whether these synergistic interactions between obesity and metabolic dysfunction are consistent in the risk of cancer mortality.
Under the Korean National Health Insurance Act, the Korean National Health Insurance system is required to cover the country's entire population.This system has provided a medical health check-up program to facilitate the early detection of diseases since 1999.The participants of the health check-up program consist of employerinsured, self-employed-insured and their dependents.We used this valuable database to conduct a retrospective cohort study investigating the risk of cancer mortality according to obesity status and metabolic health status.

Study design and population
The National Health Insurance Corporation compiled a sample database that included information regarding medical insurance, hospitals, medical health check-ups and medical treatments from 2002 to 2010.The sample cohort consisted of approximately 1 million individuals.Of these, 424,712 participants who received the health check-up program from 2002 to 2008 were assessed for eligibility for this study.The identities of the participants were removed to maintain anonymity.
Of the 424,712 subjects, 409,402 (96.4%) met the eligibility criteria, and 15,310 cases (3.6%) were excluded for the following reasons: 27 cases did not match to a death record or have dates of death, 448 died within 6 months of the health check-up, 14,574 were under 20 years of age, and 261 did not include a measurement for body mass index (BMI).Therefore, the total number of people who were eligible for the study was 409,402.We further excluded 45,521 participants who did not have covariate information.Finally, 363,881 participants were included in the analysis (Figure 1).

Data collection
The National Health Insurance-sampled cohort data included information provided by a questionnaire, anthropometric measurements and laboratory measurements, and the data also provided individual ages within a 5-year range (i.e., 20-24, 25-29, …, 75-79, ≥80 years).Data on smoking status, alcohol intake and physical activity were collected by self-reported questionnaires.Questions about alcohol intake included the frequency of alcohol consumption and the usual amount that was consumed on a daily basis.Smoking status was categorized as never smoker, former smoker or current smoker.The participants were asked about their frequency of physical activity (0, 1-2, 3-4, ≥5 times/ week).The BMI (kg/m 2 ) was calculated as weight (kg) divided by the square of the height (m).Systolic and diastolic blood pressures (mmHg) were measured.Blood samples were collected, and fasting blood glucose (mg/ dL), total cholesterol level (mg/dL) and urinary protein were measured.

Certification of death from cancer
The National Health Insurance cohort data are linked to vital statistics from Statistics Korea (Statistics Korea, 2014).Therefore, we were able to follow up for mortality and cause of death for individuals who died prior to December 31, 2010.The underlying cause of death was coded according to the International Classification of Diseases, 10 th edition (ICD-10;World Health Organization, 1994).Death from cancer was the primary end point for this study.To evaluate the risk of cancer mortality according to the metabolic health status and degree of obesity, we defined the cause of death from cancer according to the ICD-10 codes C00-C97 and D45-D47 (World Health Organization, 1994).

Statistical analysis
Data were expressed as means±SD for continuous variables, and categorical data were expressed as numbers (percentages).An independent t-test was used to compare the differences between the metabolically healthy group and the metabolically unhealthy group.A one-way ANOVA was used to compare the baseline characteristics by the obesity status.
Because the date of examination was not included in the individual records, the date of the health check-up was assumed to be the year midpoint (July 1).All participants were followed up to the date of death or December 31, 2010 via the vital statistics provided by Statistics Korea.Cox proportional hazard model analyses were used to evaluate the risk of cancer mortality according to the obesity status and metabolic health status.Model 1 was adjusted for age, sex, systolic blood pressure, diastolic blood pressure, total cholesterol level and urinary protein.
Model 2 was adjusted for Model 1 plus smoking status, alcohol intake and physical activity.We also investigated the interaction between obesity status and metabolic health status on the risk of cancer mortality using a likelihood ratio test comparing the main model (obesity status and metabolic health status) and the full model including the interaction term.p<0.05 were considered statistically significant.Statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC, USA).

Results
Table 1 shows the baseline characteristics of the study participants.During an average of 6.07 personyears of follow-up, a total of 3,391 (0.9%) individuals died from cancer.Of the 363,881 participants, 112,133 (30.8%) were obese, 87,627 (24.1%) were overweight and 164,121 (45.1%) were of normal weight.Additionally, 342,442 (94.1%) were metabolically healthy, and 21,439 (5.9%) were metabolically unhealthy.The obese group showed higher BMI, fasting blood glucose, systolic blood pressure, diastolic blood pressure and total cholesterol values compared with the normal group.The metabolically unhealthy group showed higher BMI, fasting blood glucose, systolic blood pressure, diastolic blood pressure and total cholesterol values compared with the metabolically healthy group.The proportions of positive urinary protein results, current smoker and heavy alcohol drinker statuses were higher in the obese group and the metabolically unhealthy group compared with their counterparts.
Table 2 shows the HRs and 95% CIs for risk of cancer mortality according to the obesity status and metabolic health status.According to the obesity status, after adjusting for age, sex, systolic blood pressure, diastolic blood pressure, total cholesterol level, urinary protein, smoking status, alcohol intake and physical activity, the adjusted HRs (95% CI) were 0.82 (0.75-0.89) and 0.79 (0.72-0.85) for the overweight and obese groups, respectively, compared with the normal weight group.In terms of the metabolic health status, after adjusting for age, sex, systolic blood pressure, diastolic blood pressure, total cholesterol level, urinary protein, smoking status, alcohol intake and physical activity, the adjusted HR (95% CI) was 1.26 (1.14-1.40)for the metabolically unhealthy group compared with the metabolically healthy group.
According to the obesity status, the metabolically unhealthy group had a significantly higher risk of cancer mortality compared with the metabolically healthy group after adjusting for age, sex, systolic blood pressure, diastolic blood pressure, total cholesterol level, urinary protein, smoking status, alcohol intake and physical activity (Table 3).In addition, the interaction between    obesity status and metabolic health status on the risk of cancer mortality was not statistically significant (p=0.31).

Discussion
We found that the risk of cancer mortality decreased with increasing obesity status.Conversely, a meta-analysis by McGee and the Nurses' Health Study that compared obese and normal weight subjects found that the risk of death from cancer was increased in overweight subjects (Calle et al., 1999;Calle et al., 2003;McGee et al., 2005).Additionally, the large, prospective, Cancer Prevention Study II reported 52% and 62% higher cancer mortality rates among obese males and females, respectively.The distribution of deaths according to the type of tumor differed between Korean and American populations.Further, the association between BMI and mortality is influenced by racial differences (Dahlberg et al., 2013).Nevertheless, a previous Korean study reported a slightly increased risk of death from cancer among overweight individuals (Jee et al., 2006).Mortality caused by some cancer types, such as colorectal, liver, and esophageal, increases with obesity, whereas other cancers, such as lung, show increased mortality in underweight subjects (Seidell et al., 1996;Dahlberg et al., 2013).We did not estimate the cancer mortality risk according to the type of cancer, and therefore did not distinguish cancers as obesity-related or-unrelated.
We found that the metabolically unhealthy group had a significantly higher risk of cancer mortality compared with metabolically healthy group in all obesity status groups.These findings are in accordance with those of the Framingham study, which reported that subjects with elevated blood glucose levels had higher incidences of cancer (Moore et al., 2014).Although the mechanism behind the association between glucose metabolism and cancer is not well understood, diabetes mellitus is reported to be associated with an increased risk of different types of cancer, specifically, pancreatic, liver and endometrial (relative risk range, 2.0-2.5), and breast, colon and bladder cancers (relative risk range, 1.2-1.5)(Giovannucci et al., 2010).
It is hypothesized that the mechanism behind the glucose-cancer link is related to metabolic dysfunction and inflammation (Garcia-Jimenez et al., 2013).Glucose is an essential nutrient for actively proliferating cells, and tumor cells exhibit particularly rapid growth and proliferation rates.Therefore, it is reasonable to infer that tumor cells require greater glucose uptake compared with normal cells to accelerate their proliferation.In vivo and animal studies have shown that hyperglycemia promotes tumor growth and proliferation (Nunez et al., 2006;Liu et al., 2011).Indeed, this is the basis for Positron Emission Tomography/Computed Tomography (PET/CT), which enables detection of potentially cancerous cells based on their greater uptake of glucose (Gambhir, 2002;Uzel et al., 2013).A prospective cohort study in Sweden also reported that abnormally elevated fasting glucose and postload glucose levels are associated with an increased risk of cancer incidence in females (Stattin et al., 2007).This is the first study to use sampled cohort data from a national health insurance system to examine the effects of obesity and metabolic status on the risk of cancer mortality.It demonstrated that such sampled cohort data can be representative of the entire national health insurance system.Thus, the results of this study could provide an indication of cancer risk and mortality in the Korean population.
There are a number of limitations to this study.First, we used a simplistic measure of metabolic dysfunction (fasting blood glucose) and simple anthropometric measures of body fat (BMI) to assess metabolic health and obesity, respectively.More precise measures of body fat composition and metabolic health would be useful.However, any error introduced using these simple measures was most likely non-differential.Therefore, our results are likely underestimates of the true effects.Second, we defined the endpoint simply as death due to cancer.Future studies should assess the types of cancer to evaluate obesity-related cancers.Third, the date of the health check-up was assumed to be the year midpoint (July 1), because the date of examination was not included in the individual records.This assumption may cause a measurement bias.Finally, only participants of the health check-up programs of the National Health Insurance system were assessed for eligibility.Thus, we were not able to evaluate subjects who did not participate in the health check-up programs.In addition, most participants who received the health check-up program are likely to be healthier compared with non-participants.Therefore, our sampled cohort data may be biased by the healthy volunteer effect.
In conclusion, we found that the risk of cancer mortality decreased with increasing obesity status and increased with reduced metabolic health.Therefore, given that the rates of metabolic dysfunction are rising, it can be expected that mortality due to cancer will also increase.It thus follows that treatment strategies targeting metabolic dysfunction may contribute to reducing the risk of death from cancers.

Figure 1 .
Figure 1.Flow-Chart for Study Participants Selection

Table 1 . Baseline Characteristics of Study Participants according to the Obesity Status and Metabolically Healthy Status (n=363,881)
Data are expressed as means±SD for the continuous variables and number (percentages) for the categorical variables; *p-value was tested using t-test or 1-way ANOVA for the continuous variables and chi-square test for the categorical variables.

Table 2 . Hazard Ratios (HRs) and 95% Confidence Intervals (CIs) for Risk of Death from Cancer according to the Obesity Status and Metabolically Healthy Status (n=363,881)
*Model 1 was adjusted for age, sex, systolic blood pressure, diastolic blood pressure, total cholesterol level and urinary protein; Model 2 was adjusted for model 1 plus smoking status, alcohol intake, physical activity.

Table 3 . Effects of Metabolically Unhealthy Status on the Risk of Cancer Mortality according to the Obesity Status (n=363,881)
(95% CI)*=Hazard ratios (95% Confidence Interval); *Hazard ratios and their confidence intervals are estimated using Cox proportional model after adjusting for age, sex, systolic blood pressure, diastolic blood pressure, total cholesterol level, urinary protein, smoking status, alcohol intake and physical activity; †P for interaction was tested by likelihood ratio test comparing main model and full model with all covariates DOI:http://dx.doi.org/10.7314/APJCP.2014.15.22.10027Risk of Cancer Mortality by Metabolic Health and Obesity HR