The Association of Metabolic Syndrome and Its Components with Serum Prostate-Specific Antigen Levels in a Korean-Screened Population

Prostate cancer is the most common nonskin cancer in the U.S. men. An estimated 186,320 men received a new diagnosis of prostate cancer in 2008 (1). The serum prostate-specific antigen (PSA) test is commonly used as a screening tool to rule out prostate cancer. A randomized European study showed that PSA-based screening reduced the rate of death from prostate cancer by 20% (2). Approximately 1.5 million American men ages 40 to 69 years have PSA levels of >4.0 ng/mL, a widely used cutoff value for a positive screening result (3). However, many men with “normal” PSA values have cancer, raising concern about the lack of specificity of the PSA test. Consequently, many have proposed PSA modifications in an attempt to bolster the performance of this analyte, and the search for improved biomarkers of detection has recently yielded several viable candidates such as prostate cancer gene 3, early prostate cancer antigen, promoter hypermethylation, and gene fusions (4). Researchers have also attempted to improve the accuracy of PSA test interpretation by controlling for factors such as age, prostate volume, and ethnicity (5). As a part of this strategy, several studies have examined the association between various metabolic risk factors and PSA level (6-14). Findings from these studies have urged urologists, as well as general practitioners who perform PSA tests for prostate cancer screening, not to assign the same PSA cutoffs when deciding whether a patient needs to undergo prostate biopsy.

Metabolic syndrome is a cluster of risk factors for type 2 diabetes and cardiovascular disease, and is associated with increased mortality from these and other conditions (15, 16). The Third National Health and Nutrition Examination Survey (NHANES) found that the prevalence of metabolic syndrome in the United States increased with age from 6.7% among participants ages 20 to 29 years, to 43.5% for participants ages 60 to 69 years, and 42.0% for those ages 70 years or older (17).

Metabolic syndrome is thought to play a role in the etiology of prostate cancer, although current epidemiologic evidence is not sufficient to establish a link between the two conditions (18). One prospective study from the United States found an inverse association between metabolic syndrome and the risk of developing prostate cancer (19). However, a positive association between metabolic syndrome and prostate cancer was observed in two other prospective studies and one longitudinal nested case-control study conducted in Europe (20-22). In addition, a community-based case-control study of African-Americans reported a 90% increased risk of prostate cancer associated with metabolic syndrome (23). In benign prostatic hyperplasia patients, it has been reported that the presence of metabolic syndrome was associated with an increase in prostatic growth (24, 25).

Most components of metabolic syndrome have been linked to the metabolism of testosterone. Furthermore, low serum testosterone levels are associated with obesity, insulin resistance, type 2 diabetes, and metabolic syndrome (26-33). Therefore, serum PSA level might be affected by the presence of an individual metabolic risk factor, as well as the development of metabolic syndrome.

To date, relationships between each metabolic risk factor and serum PSA levels have been actively investigated and several studies have examined the association between metabolic syndrome and serum testosterone or prostate cancer. However, few studies have addressed the possibility of a direct relationship between metabolic syndrome, a cluster of metabolic risk factors, and serum PSA levels in a screened population. To the best of our knowledge, data from the only other study to directly examine this question were not analyzed after adjustment for demographic variables, and the sample size was small (34). Although a recently published study from Korea also address the same issue, central obesity defined by waist circumference (WC) as one of the criteria for metabolic syndrome was not considered in their study due to lack of information (7). Therefore, it remains unclear whether metabolic syndrome truly affects serum PSA level in a screened population. In this study, we investigated the association of metabolic syndrome and its metabolic components with serum PSA levels in a large screened population.

Data were obtained from 26,726 men ages 40 y or older who came to the Health Promotion Center at the Asan Medical Center for a routine health checkup between January 2006 and December 2007. The health screening program available at our Health Promotion Center includes anthropometric measurements (height, weight, and WC), a blood test (a complete blood cell count, basic chemistry, serologic test, blood coagulation test, thyroid function test, and assay for tumor markers), stool/urine analysis, abdominal ultrasonography, gastrofiberscopy, chest radiography, a pulmonary function test, electrocardiography, and a detailed clinical examination. All subjects were also asked to complete a questionnaire designed to assess sociodemographic factors, comorbidities, and current or past medications. Because the Korean government has provided health insurance for the entire nation and is maintaining the policy of tightened cost controls (cost containment–centered government policy) along with a low-cost medical service system, all clinical and laboratory exams included in our “general health screening” program became available to men willing to receive the screening at a cost of ∼$1,000. The Institutional Review Board of the Asan Medical Center approved all the procedures involved in sampling and data collection, and written informed consent was obtained from all participants.

Among the 26,726 men recruited, those with a serum PSA level of 4 ng/mL or higher (1.3%, n = 339) were excluded from the study, as prostate biopsy was recommended when elevated PSA occurred in the absence of any specific condition. Subjects with a history of medication for benign prostatic hyperplasia (n = 1,622) or prostate surgery (n = 494) were also excluded (those with a history of medication for benign prostatic hyperplasia were excluded because 5-α-reductase inhibitors could cause a decrease in PSA levels). In addition, 670 men showing microscopic pyuria on urinalysis were excluded from the study. Therefore, a final population of 23,601 men was studied to examine the association of metabolic syndrome and its components with serum PSA level (Fig. 1).

Metabolic syndrome was defined using the criteria established by the National Cholesterol Education Program-Adult Treatment Panel III-American Heart Association/National Heart, Lung, and Blood Institute (NCEP-ATPIII-AHA/NHLBI) statement, published in 2005 (35). The diagnosis of metabolic syndrome was made when at least three of the following criteria were present: WC of ≥90 cm, triglyceride levels of ≥150 mg/dL or undergoing treatment for hypertriglyceridemia, high-density lipoprotein cholesterol (HDL-C) levels of <40 mg/dL or undergoing treatment for low HDL-C, blood pressure (BP) of ≥130/85 mm Hg or undergoing treatment for hypertension, and fasting plasma glucose (FPG) of ≥100 mg/dL or undergoing treatment for hyperglycemia.

All parameters were measured on fresh serum obtained after a 12-h overnight fast, when the patient had been sedentary in a sitting or supine position for 15 min. Serum total cholesterol, triglycerides, HDL-C, and FPG were measured by enzymatic methods using a chemistry autoanalyzer (Toshiba 200-FR). Low-density lipoprotein cholesterol (LDL-C) was calculated using Friedewald's formula: LDL-C = Total cholesterol − (triglyceride/5) − HDL-C. Serum PSA values were measured using a PSA-RIACT assay system (CIS Bio International).

Weight and height were measured using standard techniques. Body mass index (BMI) was calculated as weight (in kilograms) divided by the square of height (in square meters). We stratified BMI according to the recommendations of the steering committee of the Western Region of the WHO (36). Patients with BMI values of <23 kg/m² were considered underweight or normal weight; patients with BMIs of at least 23 kg/m² but <25 kg/m² were considered overweight (at risk); patients with BMIs of at least 25 kg/m² but <30.0 kg/m² were considered obese I; and patients with BMIs of 30.0 kg/m² or greater were considered obese II. A nonelastic plastic tape was used to measure WC after a normal exhalation, at the midpoint between the lower border of the rib cage and the iliac crest.

Because of the highly skewed distribution, PSA level was analyzed in continuous terms after logarithmic transformation and results were presented as geometric means with 95% confidence intervals (95% CI) to aid in the interpretation of tables. The categories of metabolic syndrome were consistent with the criteria established by the NCEP-ATP III-AHA/NHLBI statement. All subjects were dichotomized by the presence of each of five metabolic components. Associations between each of these metabolic components, as well as metabolic syndrome and serum PSA level, were assessed. Geometric mean PSA levels were compared in men with and without metabolic syndrome, and with and without each metabolic component, using Student's t test. Adjusted geometric mean PSA levels were also compared using multiple linear regression analysis.

To analyze the change in adjusted geometric mean PSA level in terms of the extent of central obesity, study subjects were further categorized into three subgroups according to the 90- and 102 cm cutoff points of WC, which are commonly used in Asian and Western populations (35). To more specifically examine the relationship between FPG levels and serum PSA levels, subjects were classified into four subgroups according to the definition established by the American Diabetes Association (normal fasting glucose: FPG, <100 mg/dL; impaired fasting glucose: FPG, 100-125 mg/dL; diabetes: FPG, ≥126 mg/dL; and men taking diabetes medication; ref. 37). Lastly, subjects were further divided into four subgroups according to measured BP, consistent with the JNC 7 guidelines (normal: <120 mm Hg systolic BP and <80 mm Hg diastolic BP; prehypertension: 120-139 mm Hg systolic BP or 80-90 mm Hg diastolic BP; hypertension: ≥140 mm Hg systolic BP or ≥90 mm Hg diastolic BP; and men taking antihypertensive medication) to measure the differences in adjusted geometric mean PSA levels in these subgroups (38). Adjusted mean PSA values were compared using one-way ANOVA and all statistical analyses were done using Stata version 10.1 (StataCorp).

The baseline demographic characteristics of the study sample are shown in Table 1. Overall, 80% of men were between 40 and 59 years of age, and 40% were obese (BMI, ≥25.0 kg/m²). The geometric mean PSA concentration was 0.66 ng/mL, and 94% of men had PSA levels of <2.0 ng/mL. The baseline metabolic characteristics of the 23,601 subjects in this cohort are shown in Table 2. The prevalence of metabolic syndrome in the entire cohort was 35.3% (8,325 of 23,601). Overall, subjects with metabolic syndrome were older than subjects without the condition. The criterion for low HDL-C was fulfilled in 44% of subjects and was the most common of the five metabolic syndrome components. About 85% of subjects fulfilled at least one criterion of metabolic syndrome (Table 3). In subjects with metabolic syndrome (n = 8,325), the most frequent metabolic syndrome component was low HDL-C (89.8%), followed by hypertriglyceridemia (74.8%), hyperglycemia (72.4%), hypertension (65.7%), and increased WC (53.1%).

Table 4 shows the associations of demographic and metabolic characteristics with the geometric and adjusted geometric mean PSA. The geometric mean PSA level increased linearly and significantly with increasing age and decreasing BMI (P < 0.001). Although the geometric mean PSA levels were lower in men with metabolic syndrome than in men without the condition, these data did not remain statistically significant after the adjustment for other demographic confounding variables such as age and BMI. Of the five metabolic components, WC (P = 0.021), BP (P < 0.001), and FPG (P = 0.001) were associated with serum PSA level, after adjusting for age, BMI, and other metabolic components. Men with increased WC had PSA values 3.2% lower than men with decreased WC [0.639 ng/mL (95% CI, 0.625-0.654) for men with a WC of ≥90 cm versus 0.660 ng/mL (95% CI, 0.652-0.675) for men with a WC of <90 cm]. Subjects with elevated FPG had PSA levels 3.5% lower than men with normal FPG [0.642 ng/mL (95% CI, 0.632-0.652) for men with a FPG of ≥100 mg/dL versus 0.665 ng/mL (95% CI, 0.656-0.674) for men with a FPG of <100 mg/dL]. However, subjects with hypertension had PSA levels 4.7% higher than men with normal BP [0.674 ng/mL (95% CI, 0.663-0.685) for men with a BP of ≥130/85 mm Hg versus 0.644 ng/mL (95% CI, 0.635-0.652) for men with a BP of <130/85 mm Hg].

Finally, we examined specific associations of PSA level with WC, FPG, and BP. In these analyses, subjects were classified into three or four subgroups by WC, FPG, and BP levels, rather than undergoing analysis as a single dichotomous group. Figure 2 shows the adjusted geometric means and 95% CIs of serum PSA levels by WC, FPG, and BP. Even after fully adjusting for age, BMI, and other metabolic risk factors, the geometric mean PSA linearly and significantly decreased as WC and FPG increased (P = 0.013 and P < 0.001; Fig. 2A and B). Subjects taking diabetic medication had the lowest PSA levels of the four subgroups. Men with increased WC (≥102 cm) and who were taking diabetic medication had PSA levels 8% (0.05 ng/mL) and 13% (0.09 ng/mL) lower than a reference group of men with decreased WC (<90 cm) and normal FPG (<100 mg/dL), respectively. However, subjects with prehypertension, hypertension, or are taking antihypertensive medication had a PSA level 5% higher (0.03 ng/mL) than a reference group of men with normal BP (P = 0.002, Fig. 2C). No significant difference in serum PSA was observed among subjects with abnormal BP (prehypertension and hypertension) and subjects who were taking antihypertensive medication (P > 0.5).

In our large screened population, serum PSA, the most widely used marker for the early detection of prostate cancer, was not associated with metabolic syndrome. Serum PSA levels were significantly associated with three of five metabolic syndrome factors defined according to the NCEP-ATPIII-AHA/NHLBI statement (35). Among metabolic risk factors, WC and FPG were inversely correlated with serum PSA, whereas BP was positively correlated. These associations remained significant after adjusting for age, BMI, and other metabolic risk factors. To our knowledge, only one prior study has directly addressed the association of metabolic syndrome with serum PSA level. This report found that metabolic syndrome was associated with decreased serum PSA; however, the sample size was small and the statistical analysis did not adjust for confounding variables that may affect serum PSA level, such as age and BMI (34). The lack of association between metabolic syndrome and serum PSA may reflect the heterogeneous relationship between each metabolic risk factor and serum PSA (WC and FPG were negatively correlated, whereas BP was positively correlated, with serum PSA levels). Although metabolic syndrome is considered an emerging hypothesis in the etiology of prostate cancer, and some metabolic risk factors are associated with serum PSA, our results suggest that physicians should not independently consider this highly prevalent disease entity as a confounding variable in the interpretation of serum PSA level.

It has been reported that PSA levels are lower in obese men than in men of normal weight because of the obesity-related hemodilution or the decrease in circulating androgens found in obese men (8-10, 39, 40). We further confirmed the association of obesity and a lower serum PSA level in a screened Asian population, and our data suggest that obesity may affect the interpretation of the serum PSA level. In our study, central obesity as determined by WC was independently associated with serum PSA level even after adjusting for age and BMI. WC is thought to be the best anthropometric measure of total body fat, as well as the best simple indicator of intra-abdominal fat mass (41). The inverse relationship between PSA level and WC observed in our study are contradictory to previously reported results. Rundle et al. (42) found that WC was positively associated with PSA level, even after controlling for BMI. However, they did not jointly analyze confounding variables (significant metabolic risk factors affecting serum PSA level, such as diabetes and hypertension). We found that men with a WC of ≥90 cm had significantly lower levels of serum PSA, compared with those whose WC was <90 cm, even after adjusting for BMI as a surrogate of overall body fatness. Furthermore, we observed a linear negative relationship between WC and serum PSA levels, whereby increased WC was associated with lower mean PSA level. Previous studies have reported that WC is a stronger predictor than BMI of low endogenous testosterone level, especially with regards to free testosterone (43, 44). Therefore, the inverse association between WC and serum PSA level, which was maintained after fully adjusting for age, overall adiposity, and other metabolic components, might reflect strong underlying associations between abdominal obesity and endogenous testosterone levels.

Previous studies have shown that men with diabetes have significantly lower serum testosterone concentrations than nondiabetic men (45-48). In our study, subjects with high FPG had lower serum PSA levels than men with normal FPG. In addition, a significant inverse linear relationship was observed between FPG and serum PSA levels. Men taking medication for diabetes had serum PSA levels 13% lower than subjects with normal FPG (<100 mg/dL). The observed association between diabetes and serum PSA level is consistent with previous observations that men with type 2 diabetes have lower serum PSA levels than healthy subjects (13, 14, 49).

Also, significantly, we found that men with high BP had higher serum PSA levels than men with normal BP even after fully adjusting for confounding variables such as age, BMI, and other metabolic factors. To date, the association of high BP with serum PSA levels remains controversial. Han et al. (7) reported that elevated diastolic BP, but not systolic BP, was associated with elevated serum PSA. In contrary, a NHANES study conducted in the United States found that high BP was not associated with serum PSA level (14). In our study, men who had abnormally elevated BP or who were taking antihypertensive medication had higher serum PSA levels than subjects with normal BP. However, no difference in serum PSA level was observed among the three subgroups (men with prehypertension, men with hypertension, and men taking antihypertensive medication). The conflicting relationship reported in the literature might have resulted from the differences in study sample size, the inclusion criteria for study subjects (age and ethnicity), and the definition of hypertension (cutoff point of hypertension and information about antihypertensive medication). Our study used the definition of the JNC-7 guideline, which is more commonly used, as well as updated guidelines about the diagnosis of hypertension. Most importantly, as information about antihypertensive medications was available in our study subjects, and subgroup analysis could be done according to the JNC 7 guidelines to investigate the association of hypertension with serum PSA. Although the exact mechanisms are unknown, the positive relationship observed in our study between hypertension and PSA level possibly resulted from long-term exposure to androgens, which may trigger multiple vasoconstricting mechanisms. Experimental studies have shown that androgens upregulate thromboxane A2 expression, norepinephrine synthesis, neuropeptide Y, angiotensin II expression, and endothelin-1 action (50). Future research is needed to explore the potential associations between BP elevation and prostate cancer diagnosis in detail, as well as serum PSA levels in healthy men, especially with regard to the effects of serum testosterone.

Although the relationship between circulating androgens and serum PSA level in men without prostate cancer has not been clearly established, a recent study from the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial showed that PSA correlates positively with total and free serum testosterone levels in men at low risk for prostate cancer (has serum PSA of <4 ng/mL and has not been diagnosed with prostate cancer through 5 years of follow-up; ref. 51). Therefore, PSA changes according to the metabolic profiles observed in our study might be partially explained by the differences in underlying testosterone level because previous studies showed the differences in circulating androgens according to the presence of metabolic risk factors.

The geometric mean PSA levels in our study population were 19% and 29% lower than values observed in the U.S. NHANES population in men ages 40 to 49 years and 50 to 60 years, respectively (14). These findings may reflect ethnic differences between Western and Asian populations. Previous studies have reported that screened Asian populations have lower PSA levels than Western populations of similar age ranges (52). Therefore, the magnitude of PSA change according to the metabolic risk factors observed in our study might be small compared with Western populations. These findings should be further investigated in a Western screening population as well as in men with serum PSA levels higher than those commonly used as cutoff points for prostate biopsy.

Although our study was not community based, the high prevalence of obesity (41.5%) and metabolic syndrome (35.3%) observed were similar to the results from Korean NHANES data (the prevalence of obesity and metabolic syndrome were 32.9% and 30.7% in Korean adults ages at least 20 years according to the criteria of obesity for Asians and NCEP-ATP III-AHA/NHLBI metabolic syndrome definition; ref. 53). Considering the recent marked increase in the number of obese men in Korea and older cohort population compared with those from Korean NHANES, it was presumed that the difference between the prevalence of obesity and metabolic syndrome in our study population and those from community based study was quite small.

Our study was strengthened by the large size of the screened cohort population, the use of standardized clinical and laboratory covariates, and the clearly documented questionnaire. In addition, all subjects were 40 years of age or older, consistent with the age when regular PSA screening for early detection of prostate cancer is initiated. However, our study was limited by our inability to measure and analyze endogenous testosterone levels as well as plasma volume, both of which might have more clearly elucidated the underlying links between metabolic risk factors and serum PSA levels. In addition, it was difficult to analyze the duration of any metabolic risk factor as a covariate, because a substantial number of patients with metabolic risk factors may be undiagnosed and the duration of such risk factors may reflect the extent of medical surveillance. Lastly, because digital rectal examination was not included as part of our prostate cancer screening program and prostate cancer was not ruled out by prostate biopsy, we cannot definitely rule out the presence of undiagnosed prostate cancer.

Our findings suggest that clinical interpretation of serum PSA levels in men with metabolic risk factors may be biased. However, the magnitude of the PSA change according to the metabolic risk factors observed in our study may seem small even when the ethnic differences between Western and Asian populations are considered. Therefore, it remains unclear whether the presence of a metabolic risk profile could affect the delay in the detection of prostate cancer. Further prospective studies are needed to identify any potential delay in the detection of prostate cancer, considering the duration and severity of the metabolic risk factors in other races. In addition, based on our results, further studies are also needed to assess how the severity and duration of each metabolic risk profile in each individual affects the serum PSA levels. Also, because the clinical effect of metabolic risk factors on prostate cancer detection likely extends beyond the serum PSA level, the relationship between the metabolic risk factors and other confounding factors such as PSA testing practices and biopsy referral should be further investigated.

In summary, our data show that metabolic syndrome is not associated with serum PSA levels in a screened population. Among the risk factors for metabolic syndrome, BP, WC, and FPG were independently associated with serum PSA levels. The lack of association between metabolic syndrome and serum PSA levels may reflect the heterogeneous relationship between each metabolic risk factor and serum PSA. Our study suggests that the clinical interpretation of serum PSA levels may be biased by these metabolic risk factors, as well as by age and BMI, but not by metabolic syndrome by itself. Given the prevalence of these risk factors, the association of each metabolic risk factor with PSA could have important ramifications for cancer detection and should be further quantified.

No potential conflicts of interest were disclosed.

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.