The association of body mass index and prostate-specific antigen in a population-based study




Recent studies of men with prostate carcinoma suggest that obesity may be associated with more advanced-stage disease and lower overall survival rates. One possible link between body mass index (BMI) and prostate carcinoma prognosis may be disease ascertainment. Prostate-specific antigen (PSA) is widely used to screen for prostate carcinoma.


The authors examined the association between BMI and PSA in a population-based study of 2779 men without prostate carcinoma. Between 2001 and 2004, these men were enrolled in a study sponsored by the San Antonio Center of Biomarkers of Risk, a clinical and epidemiologic center of the Early Detection Research Network of the National Cancer Institute.


The mean PSA value decreased in a linear fashion with an increase in BMI category, from 1.01 ng/mL in normal weight men to 0.69 ng/mL in obese (Class III) men, after adjusting for race/ethnicity and age.


Lower levels of PSA in obese and overweight men could mask biologically consequential prostate carcinoma. Cancer 2005. Published 2005 by the American Cancer Society.

Obesity is a major public health problem in the U.S. and is linked to chronic diseases such as cancer.1, 2 Although some studies report an increased risk of prostate carcinoma among obese and overweight men,3–7 others indicate little or no association.5, 8–10 More consistent findings have linked obesity with prostate carcinoma mortality, advanced stage, and higher Gleason grade.5–7 The mechanisms whereby obesity may lead to these adverse outcomes include increased levels of insulin-like growth factor and estrogens as well as decreased levels of sex hormone binding globulin.6, 11, 12 One possible link between body mass index (BMI) and prostate carcinoma incidence and prognosis may be disease ascertainment. Prostate-specific antigen (PSA) is widely used to screen for prostate carcinoma and is the most common determining factor for disease diagnosis.13 Although PSA levels vary by age and ethnicity, few data exist on the association between BMI values and serum PSA levels.13 The purpose of the current investigation was to asssess the association between BMI and PSA.

The San Antonio Center for Biomarkers of Risk of prostate carcinoma (SABOR) is a Clinical and Epidemiologic Center of the Early Detection Research Network and is supported by the National Cancer Institute. The objective of SABOR is to develop new methods of prostate carcinoma detection using behavioral, genetic, and other markers of risk of the disease in a cohort that will ultimately number 10,000. Recruitment of a multiethnic, population-based sample was achieved using outreach clinics throughout metropolitan San Antonio, Texas. Healthy men without a history of prostate carcinoma are eligible for participation. After informed consent was obtained, men complete an extensive series of instruments (demographics, diet, quality of life, family history, ethnicity/race, American Urologic Association [AUA] symptom score), provide biologic samples (including serum for PSA measurement), and undergo a directed physical examination including a digital rectal examination (DRE) and height, weight, and anthropometric measures. PSA was measured using the Abbott immunoassay system (IMX) (Abbott Laboratories, Abbott Park, IL). Participant enrollment into SABOR began in March 2001 with annual follow-up examinations. A concerted effort is made to oversample ethnic minorities and medically underserved populations.

Of 2841 men who were recruited to participate in SABOR between March 2001 and 2004, 71 were found to have prostate carcinoma after a prostate biopsy was performed, prompted by an abnormal DRE result or elevated PSA level. The cohort for the current study consisted of the remaining 2770 men without prostate carcinoma. BMI (weight in kg/height in m2) was calculated from measured height and weight and was categorized as follows: normal and underweight (BMI < 24.9), overweight (BMI 25.0–30.9), obese Class I (BMI 30.0–34.9), obese Class II (BMI 35.0–39.9), and obese Class III (BMI ≥ 40.0).14

Statistical Methods

All statistical analyses were performed with the Statistical Analysis System, Version 8.0 (SAS Institute Inc., Cary, NC). Bivariate and multivariate general linear models were used to assess the independent contributions of age, race, and BMI in explaining the continuous dependent variable, PSA level. Because of its highly skewed distribution, PSA level was examined as the natural log function and subsequently back transformed for interpretation of the model results. Linear contrasts were used to examine the significance of the linear trend in mean PSA level for the rank-ordered covariates. We also examined the conditional effects of both age and race with BMI on PSA level by fitting first-order interaction terms in multivariate models and by conducting analyses stratified by age and race in contingency tables.

Table 1 shows the distributions of demographic characteristics and BMI in the study cohort, as well as the association of each of these factors with mean PSA value. The mean age of the study cohort was 56.7 years (standard deviation [SD] 10.8). Of the enrollees, 50.1% were Non-Hispanic white, 36.5% were Hispanic, and 12.4% were African American. Overall, 81.4% of the cohort was classified as either overweight or obese. The mean PSA value for the entire study cohort was 1.3 ng/mL (SD 1.4).

Table 1. PSA Levels by Cohort Characteristics
Mean PSA (SD)
Cohort characteristicsNo. of patients (%) RawGeometricaAdjustedb
  • PSA: prostate-specific antigen; SD: standard deviation; BMI: body mass index.

  • a

    Data were log transformed and back transformed.

  • b

    Data were log transformed and back transformed and adjusted for other study variables (i.e., age, race/ethnicity, body mass index).

Full sample27701001.32 (1.40)0.94 (0.78)
Age (yrs)
 ≥ 7033712.22.041.431.28
P for linear trend  < 0.0001< 0.0001< 0.0001
 Non-Hispanic white138950.11.391.000.82
 Hispanic white101036.51.290.900.85
 African American34312.41.100.820.98
 Other   < 0.0001 
P value  0.07 0.28
 < 24.9 (normal)51918.71.421.031.01
 25.0–29.9 (overweight)1,31847.61.360.970.95
 30.0–34.9 (obese Class I)65923.81.240.900.91
 35.0–39.9 (obese Class II)1906.91.150.760.81
 ≥ 40 (obese Class III)843.00.940.630.69
P for linear trend  < 0.0005< 0.0001< 0.0001

Adjusted and unadjusted mean values of PSA are presented according to each of the study variables of interest (Table 1). The bivariate models show that both age and BMI exhibit strong linear associations with PSA level. PSA level increased in a monotonic fashion according to age category, and decreased monontically according to BMI category. The bivariate model also indicates a significant effect for race/ethnicity. Pairwise comparisons indicate that this association is driven primarily by the differences between Non-Hispanic whites versus Hispanics and by differences between Non-Hispanic whites versus African Americans. In the adjusted model, only the effects of age and BMI remained statistically significant. Both variables exhibited strong linear trends consistent with those shown in the bivariate models. Figure 1 shows the least-square means and 95% confidence intervals of PSA according to BMI category, after adjusting for age and race/ethnicity. There was no evidence of a differential effect of BMI on PSA level across race or age categories. Trends for decreasing PSA level with increasing obesity were similar in analyses stratified by both age and race. Consistent with this, our examination of interaction terms in the multivariate model showed no evidence of effect modification of BMI on PSA level by either race (P = 0.73) or age (P = 0.76).

Figure 1.

Adjusted least-squares mean prostate-specific antigen (PSA) according to body mass index (BMI) category. 95% CI: 95% confidence interval.

Recent studies suggest that obesity may be associated with more advanced disease and lower overall survival rates in men with prostate carcinoma.5–7 Results of the Prostate Cancer Prevention Trial (PCPT) have highlighted ascertainment bias in prostate carcinoma detection, finding worrisome rates of prostate carcinoma in men at all levels of PSA.15 As PSA is the primary reason for prostate biopsy and subsequent prostate carcinoma diagnosis, it is important to determine any impact other factors may have on PSA levels. Others have reported the relation between PSA and age/ethnicity but little is known about the association with BMI.13

We found that BMI exhibits a strong linear association with PSA (P < 0.02), which persists after adjustment for age and race/ethnicity. A substantial decrease in PSA level occurred with each increase in BMI category. Of the possible explanations for this observation, it is intuitive that lower circulating levels androgens and increased levels of estrogens in obese men could affect PSA production due to the androgen response element in the PSA promoter region.16, 17

The implications of this observation, as well as the recent reports of the PCPT showing the presence of high-grade cancer in men with levels of PSA ≤ 4.0 ng/mL, call into question whether the recent reports of inferior outcomes of prostate carcinoma treatment in obese men may be caused by delayed detection in obese men.6, 18 Given the increased rates of obesity in the U.S.,1, 2 understanding the specific pathways by which obesity is associated with prostate carcinoma incidence and disease ascertainment is critically important. Future investigations should assess whether consideration of BMI in PSA interpretation results in earlier detection and, ultimately, in better prostate carcinoma prognosis for obese men.