Abstract
Associations of serum vitamin A and carotenoid levels with markers of prostate cancer detection were evaluated among 3927 U.S. men, 40–85 years of age, who participated in the 2001–2006 National Health and Nutrition Examination Surveys. Five recommended definitions of prostate cancer detection were adopted using total and free prostate specific antigen (tPSA and fPSA) laboratory measurements. Men were identified as high-risk based on alternative cut-offs, namely, tPSA>10 ng/ml, tPSA>4 ng/ml, tPSA> 2.5 ng/ml, %fPSA<25% and %fPSA<15%. %fPSA was defined as (fPSA÷tPSA)×100%. Serum levels of vitamin A (retinol, retinyl esters) and carotenoids (α-carotene, β-carotene, β-cryptoxanthin, lutein+zeaxanthin, lycopene) were defined as quartiles and examined as risk/protective factors for PSA biomarkers. Odds ratios (OR) and 95% confidence intervals (CI) were estimated using binary logistic models. After adjustment for known demographic, socioeconomic and lifestyle confounders, high serum levels of retinyl esters (tPSA>10 ng/ml: Q4vs.Q1→OR=0.38, 95% CI: 0.14–1.00) and α-carotene (%fPSA<15%: Q4vs.Q1→OR=0.49, 95% CI: 0.32–0.76) were associated with a lower odds whereas high serum level of lycopene (tPSA>2.5 ng/ml: Q4vs.Q1→OR=1.49, 95% CI: 1.01–2.14) was associated with a greater odds of prostate cancer detection. Apart from the three significant associations observed, no other exposure-outcome association was significant. Monitoring specific antioxidant levels may be helpful in early detection of prostate cancer.
Keywords: vitamin A, carotenoids, prostate cancer, prostate-specific antigen
INTRODUCTION
The burden of prostate cancer (PCa) remains a public health concern. PCa is the most commonly diagnosed non-dermatological cancer and the second leading cause of cancer-related death among U.S. men, with an estimated 217,730 new cases and 32,050 deaths in 2010 (1). In the U.S., a sharp rise in screen-detected PCa cases has been coupled with a more gradual decline in PCa deaths over the past 20 years (2). These secular trends have been attributed not only to aging of the U.S. population but also to adoption of serum prostate-specific antigen (PSA) testing as a routine method to screen for PCa (2, 3). Despite its low sensitivity and specificity, the PSA test is twice as likely as the digital rectal exam (DRE) to identify early-stage PCa (2). The American Cancer Society recommends that men should have the opportunity to make an informed decision about whether to be screened for PCa after discussion with their health care provider, who can inform them about uncertainties, risks, and potential benefits of PCa screening; discussion should start at age 50 years for average-risk men with an estimated life expectancy of 10 years or more, but can start as early as age 40 years for higher-risk men (http://www.cancer.org/Cancer/ProstateCancer/MoreInformation/ProstateCancerEarlyDetection/prostate-cancer-early-detection-acs-recommendations). The American Urological Association recommends offering PCa testing to asymptomatic men 40 years of age or older who wish to be screened with an estimated life expectancy of more than 10 years (http://www.auanet.org/content/guidelines-and-quality-care/clinical-guidelines/main-reports/psa09.pdf).
Although the U.S. Preventive Services Task Force recognizes that evidence for or against PCa screening is insufficient (4), the PSA test is routinely performed by clinical practitioners for the early detection of PCa, and this has resulted in a greater yield of localized PCa among U.S. men (5). Total PSA (tPSA) and free PSA (fPSA) are measured and percent free PSA (percentfPSA) is calculated. Whereas tPSA threshold values (tPSA>10 ng/ml, tPSA>4.0 ng/ml or tPSA >2.5 ng/ml) for recommending a prostate biopsy are still under debate (2, 5, 6), threshold values for percentfPSA (percentfPSA<25 percent or percentfPSA<15 percent) have been proposed as additional PCa screening tools (5, 6).
Dietary antioxidants which include vitamin A and carotenoids may provide an avenue for PCa prevention. Besides age and family history, few risk factors for PCa are currently established. Whereas the etiology of PCa is largely unknown (7–16), inflammation caused by infection, hormonal changes, physical trauma, urine reflux and dietary habits, is believed to play a role in prostate carcinogenesis (7, 17–20). Micronutrients with anti-oxidative properties are primarily obtained through dietary consumption of fruits and vegetables and are capable of scavenging harmful reactive oxygen species (ROS) that are generated during inflammatory processes. In sum, the biological mechanism underlying a protective role for vitamin A, carotenoids and other antioxidants against prostate carcinogenesis may include reversal of DNA damage by ROS (7).
To date, no studies using a representative sample of the U.S. population have examined levels of vitamin A (retinol, retinyl esters) and other carotenoids (α-carotene, β-carotene, β-cryptoxanthin, lutein, zeaxanthin, lycopene) as risk/protective factors for PSA biomarkers. Epidemiological evidence for the protective effect of lycopene – a carotenoid found in tomato and tomato-based products – against screen-detected PCa has yet to be established through randomized controlled trials (21). Recent studies based on the National Health and Nutrition Examination Surveys (NHANES) have described an inverse relationship between tPSA level and body mass index (BMI) (2, 22). Higher plasma volume and lower testosterone levels are physiological alterations that may explain this relationship as well as the delayed detection, reduced incidence and worse outcome of PCa among obese U.S. men (2, 23). One study suggested that regular use of non-steroidal anti-inflammatory drugs (NSAIDs) may reduce tPSA level, a further evidence linking inflammatory processes, and ultimately antioxidants, to PCa detection among U.S. men (7). Finally, one study reported a limited influence of diet and exercise on tPSA, but did not evaluate the potential role of antioxidants (23).
The purpose of this study is to analyze the cross-sectional associations of serum vitamin A and carotenoid levels with markers of PCa detection (tPSA and percentfPSA) among U.S. men, using 2001–2006 NHANES data.
MATERIALS AND METHODS
Data source
The NHANES is a cross-sectional, nationally representative survey designed to assess the health and nutritional status of the U.S. civilian non-institutionalized population (5). The survey is unique in that it combines interviews and physical examinations. Stratified, multistage, probability survey samples were obtained based on the selection of counties, blocks, households and persons within households (5), with over-sampling of individuals with low income, adults aged 60 years or older, African-Americans, and Mexican-Americans (24). Demographic, socioeconomic and health data were collected by trained staff using household interviews. In addition, a mobile examination center (MEC) run by health professionals collected numerous measurements including anthropometric, blood pressure, and laboratory tests, either on all or a sub-group of study participants. Since 1999, NHANES has become a continuous surveillance system. For the current analyses, we combined the 2001–2002, 2003–2004 and 2005–2006 NHANES datasets. Response rates for the 2001–2002, 2003–2004, and 2005–2006 surveys were 84 percent, 79 percent and 80 percent for the interview component, and 80 percent, 76 percent and 77 percent for the physical examination component, respectively (24). Informed consent was obtained for all participants and the institutional review board of the National Center for Health Statistics, Centers for Disease Control and Prevention approved all protocols for the NHANES (25).
Inclusion and exclusion criteria
A total of 4869 2001–2006 NHANES participants were eligible for PSA testing based on gender (male) and age (≥ 40 years). Of 4270 offered PSA testing, 4166 were willing to undergo PSA testing, 102 refused and 2 did not provide a response. A total of 38 subjects had missing data on exclusion criteria and 201 subjects fulfilled at least one of the exclusion criteria (Current infection or inflammation of the prostate gland (n=50); DRE in the past week (n=60); Prostate biopsy in the past 30 days (n=10); Cytoscopy in the past 30 days (n=16); History of PCa (n=179)) (5, 7). Of the remaining 3927 subjects who satisfied inclusion/exclusion criteria, 107 had missing data on tPSA and fPSA levels. Complete-case analyses were performed on subsets of 3820 subjects with known PSA laboratory measurements. Overall, subjects with known (n=3820) and missing (n=107) PSA biomarker data were similar in terms of age, education, marital status, household income and smoking history, but differed according to race (Hispanic: 8.6% vs. 5.7%; Non-Hispanic White: 79.1% vs. 72.4%; Non-Hispanic Black: 8.8% vs. 19.1%; Other: 3.6% vs. 2.8%; P=0.01).
Laboratory measurements
Blood samples were collected by venipuncture, centrifuged, and sera were frozen at −20°C within 1 hour of phlebotomy (5). Within 1 week, frozen specimens were sent on dry ice to the University of Washington Medical Center, Department of Laboratory Medicine, Immunology Division Laboratory (Seattle, WA), where they were kept at −70°C until analyzed (5).
Hybritech tPSA and fPSA monoclonal antibody assays (Hybritech, San Diego, CA) were applied on Beckman Coulter Access analyzer (Fullerton, CA) (2, 5, 6, 25). The coefficient of variation (CV) for tPSA test was less than 4.8 percent and that of fPSA was less than 5.5 percent (5). In this study, tPSA measurements were considered abnormal if found to exceed pre-specified cutpoints, namely 2.5 ng/mL, 4.0 ng/mL or 10.0 ng/mL (2, 5, 6). Similarly, percentfPSA measurements, defined as (fPSA÷tPSA)×100 percent, were considered abnormal based on two alternative cutpoints, namely <15 percent or <25 percent (5, 6).
High performance liquid chromatography with photodiode array detection was used to measure serum concentrations of retinol, α-tocopherol, γ-tocopherol, retinyl palmitate, retinyl stearate, α-carotene, trans-β-carotene, cis β-carotene, β-cryptoxanthin, lutein+zeaxanthin, trans-lycopene and total lycopene. In this study, vitamin A was measured as retinol and retinyl esters (sum of retinyl palmitate and retinyl stearate). Additionally, α-carotene, β-carotene (cis+trans), β-cryptoxanthin, lutein+zeaxanthin, total lycopene and total carotenoids were also examined. The CVs for vitamin A and β-carotene were previously reported to be less than 5%, whereas the CVs for other carotenoids were less than 11%. Furthermore, the lower limits of detection for retinol, retinyl esters and carotenes were 0.003 µg/mL, 0.005 µg/mL and 0.003 µg/mL, respectively. Vitamin A and carotenoid measurements were analyzed in relation to tPSA and percentfPSA biomarkers, after categorization into quartiles (Q1, Q2, Q3, Q4).
Covariates
A priori confounders for the hypothesized relationships were identified from previously conducted research of similar populations of U.S. men that focused on determinants of abnormal PSA levels (7, 23, 25–27). These include age (‘40–49’, ‘50–59’, ‘60–69’, ‘70–79’, ‘80+’ years), race/ethnicity (‘Hispanic’, ‘Non-Hispanic White’, ‘Non-Hispanic Black, Other’), education (‘Less than High School’, ‘High School’, ‘More than High School’), marital status (‘Ever married’, ‘Never married’), household income (‘< $20,000’, ‘≥ $20000’) and BMI (‘<25 kg/m2’, ‘25-<30 kg/m2’, ‘≥ 30 kg/m2’). Smoking history (‘Current smoker’, ‘Former smoker’, ‘Never smoker’) was considered but not included in the final model due to sample size limitations. Other obesity-related characteristics including insulin resistance, metabolic syndrome, physical exercise and dietary intake (total energy, percent carbohydrate and percent fat) were not considered as potential confounders due to large percentages of missing data.
Statistical analysis
Descriptive, bivariate and multivariate analyses were conducted using STATA version 8. Using survey commands, we applied recommended sub-sample weights for the period of 2001–2006. MEC exam weights were used for all analyses. Masked variance units were used to estimate variances through a Taylor series linearization method. The NHANES analytic guidelines which include sub-sample and MEC exam weights can be located at the following address: http://www.cdc.gov/nchs/nhanes/nhanes2003-2004/analytical_guidelines.htm. Bivariate associations were analyzed using Pearson’s Chi-square tests for independence and one-way analysis of variance tests. Unadjusted and fully-adjusted beta coefficients and odds ratios (OR) with their 95 percent confidence intervals (CI) were computed using svylogit and svyreg commands, taking sampling weights into consideration. These weights were defined to represent the U.S. civilian, non-institutionalized population while accounting for over-sampling of certain age and ethnic groups and interview non-response. Primary analyses were based on logistic regression (dichotomous outcomes) using quartiles for the exposures; further analyses were performed using linear regression (continuous outcomes) with exposures defined as quartiles, but also examining fractional polynomials and quadratic terms (28). Specifically, we compared the odds of PCa detection between the upper and lower quartiles (Q4 vs. Q1) of vitamin A and carotenoid levels. In addition, the associations of serum vitamin A and carotenoid concentrations with tPSA (Loge transformed) and (fPSA÷tPSA)×100 percent were evaluated using least squares regression models. Sensitivity analyses were also conducted using multivariable linear regression models for log-transformed PSA biomarkers as a function of carotenoids defined as continuous variables, and quadratic terms of exposure variables were added to better characterize the non-linear associations between carotenoids and PSA biomarkers after adjustment for confounders. Sampling weights were applied in all models. Two-sided statistical tests were performed at an alpha level of 0.05.
RESULTS
The study sample consisted of 3927 U.S. male adults with a mean (SEM) age of 55.13 (0.29) years. Nearly 79 percent were non-Hispanic White, 57 percent had more than high school education, 89 percent were ever married, 86 percent had a household income of $20000 or higher, 25 percent were current smokers and 34 percent were obese (Table 1).
Table 1.
Socio-Demographic and Lifestyle Characteristics of Study Participants – 2001–2006 National Health and Nutrition Examination Survey (N=3820)*
| Sample Size | Population estimates | |
|---|---|---|
| Age (years) | N=3820 | Mean (SEM) |
| 55.09 (0.29) | ||
| N=3820 | Percent (SEM) | |
| 40–49 | 1115 | 39.4 (1.3) |
| 50–59 | 827 | 29.4 (1.2) |
| 60–69 | 840 | 16.5 (0.6) |
| 70–79 | 655 | 10.3 (0.5) |
| 80–85 | 383 | 4.1 (0.3) |
| Race and Ethnicity | N=3820 | Percent (SEM) |
| Hispanic | 805 | 8.6 (1.2) |
| Non-Hispanic White | 2205 | 79.0 (1.6) |
| Non-Hispanic Black | 709 | 8.8 (0.8) |
| Other | 101 | 3.6 (0.4) |
| Education | N=3815 | Percent (SEM) |
| Less than High School | 1178 | 17.3 (0.9) |
| High School | 876 | 25.6 (0.7) |
| More than High School | 1761 | 57.1 (1.3) |
| Marital Status | N=3816 | Percent (SEM) |
| Ever Married | 3392 | 88.6 (0.7) |
| Never Married | 424 | 11.4 (0.7) |
| Household income ($) | N=3628 | Percent (SEM) |
| < 20000 | 821 | 13.8 (0.8) |
| ≥ 20000 | 2807 | 86.2 (0.8) |
| Smoking history | N=2644 | Percent (SEM) |
| Current smoker | 641 | 24.6 (1.2) |
| Former smoker | 1069 | 37.4 (1.2) |
| Never smoker | 934 | 37.9 (1.4) |
| Body mass index (kg/m2) | N=3713 | Percent (SEM) |
| < 25 | 905 | 21.9 (0.8) |
| 25-<30 | 1621 | 43.8 (1.1) |
| 30+ | 1187 | 34.2 (1.1) |
SEM=Standard Error of the Mean;
Out of 3927 subjects, 3820 (97.3%) had known tPSA level and 107 (2.7%) had missing data on tPSA level. These two groups differed significantly according to race, but none of the other socio-demographic characteristics.
tPSA level was < 2.5 ng/mL, 2.5-<4.0 ng/mL, 4.0-<10 ng/mL and ≥ 10 ng/mL in 87.5 percent, 6.3 percent, 5.1 percent and 0.9 percent of the study sample, respectively. Whereas 65.2 percent had a percentfPSA ≥ 25.0 percent, 26.5 percent had a percentfPSA of 15-<25 percent and 8.3 percent had a percentfPSA of < 15 percent. Antioxidant concentrations ranged between 0.075 µmol/L for α-carotene and 2.32 µmol/L for retinol (Table 2).
Table 2.
Outcome and Exposure Characteristics of Study Participants – 2001–2006 National Health and Nutrition Examination Survey.
| Sample Size | Population Estimates | Median (Inter-quartile range) | |
|---|---|---|---|
| tPSA (ng/ml) | N=3820 | Mean (SEM) | |
| 1.51 (0.04) | 1.0 (0.6–1.9) | ||
| N=3820 | Percent (SEM) | -- | |
| < 2.5 | 3141 | 87.5 (0.7) | -- |
| 2.5-<4 | 323 | 6.3 (0.5) | -- |
| 4-<10 | 288 | 5.1 (0.4) | -- |
| 10+ | 68 | 0.9 (0.1) | -- |
| % fPSA | N=3820 | Mean (SEM) | |
| 30.9 (0.4) | 28.8 (21.4–38.3) | ||
| N=3820 | Percent (SEM) | -- | |
| < 15 | 355 | 8.3 (0.6) | -- |
| 15-<25 | 1044 | 26.5 (1.0) | -- |
| 25+ | 2421 | 65.2 (1.3) | -- |
| Retinol (µmol/L) | N=3823 | Mean (SEM) | |
| 2.32 (0.01) | 2.2 (1.9–2.6) | ||
| Retinyl esters (µmol/L) | N=3586 | Mean (SEM) | |
| 0.10 (0.003) | 0.07 (0.05–0.1) | ||
| α-carotene (µmol/L) | N=3817 | Mean (SEM) | |
| 0.075 (0.003) | 0.05 (0.03–0.09) | ||
| β-carotene (µmol/L) | N=3709 | Mean (SEM) | |
| 0.33 (0.009) | 0.05 (0.1–0.4) | ||
| β-cryptoxanthin (µmol/L) | N=3802 | Mean (SEM) | |
| 0.16 (0.003) | 0.2 (0.08–0.2) | ||
| lutein+zeaxanthin (µmol/L) | N=3816 | Mean (SEM) | |
| 0.29 (0.004) | 0.3 (0.2–0.4) | ||
| Lycopene (µmol/L) | N=3816 | Mean (SEM) | |
| 0.43 (0.004) | 0.4 (0.2–0.5) | ||
| Total carotenoids (µmol/L) | N=3698 | Mean (SEM) | |
| 1.28 (0.02) | 1.1 (0.8–1.6) |
%fPSA = Percent free prostate specific antigen; SEM=Standard Error of the Mean; tPSA = Total prostate specific antigen.
As shown in Table 3significant differences in log-transformed tPSA and percentfPSA levels were observed according to selected socio-demographic characteristics. When defined as dichotomous variables, tPSA levels consistently increased with age, but no similar trends were noted for percentfPSA. There were no significant racial or ethnic disparities, except for Hispanics who had the highest proportion of percentfPSA < 25 percent. Two biomarkers of PCa detection (tPSA > 2.5 ng/ml and percentfPSA < 15 percent) were inversely associated with level of education. A significantly higher proportion having tPSA > 2.5 ng/ml was found in the ever versus never married group of men. Having a household income of at least $20000 was associated with a higher prevalence of tPSA > 4 ng/mL, tPSA > 2.5 ng/mL and percentfPSA < 15 percent. Current smokers had the lowest proportion of men having tPSA > 4 ng/mL, whereas BMI was inversely related to tPSA > 4 ng/mL and tPSA > 2.5 ng/mL.
Table 3.
Serum Prostate-Specific Antigen Levels by Socio-Demographic and Lifestyle Characteristics of Study Participants – 2001–2006 National Health and Nutrition Examination Survey
| tPSA > 10 ng/ml (%) |
tPSA > 4 ng/ml (%) |
tPSA > 2.5 ng/ml (%) |
%fPSA < 25% (%) |
%fPSA < 15% (%) |
tPSA Mean (SD) |
%fPSA Mean (SD) |
|
|---|---|---|---|---|---|---|---|
| Total, Mean (SEM) | 0.99 (0.1) | 6.1 (0.4) | 12.4 (0.7) | 34.8 (1.3) | 8.3 (0.6) | ||
|
Age (years) (N=3820) |
|||||||
| 40–49 | 0.23 | 1.40 | 3.40 | 32.02 | 7.86 | 0.93 (0.90) | 31.3 (12.2) |
| 50–59 | 0.51 | 3.36 | 8.81 | 36.05 | 7.73 | 1.33 (1.97) | 30.2 (12.6) |
| 60–69 | 1.54 | 9.14 | 20.72 | 39.54 | 8.72 | 2.06 (3.55) | 29.6 (12.9) |
| 70–79 | 2.76 | 19.40 | 33.15 | 32.68 | 9.98 | 2.89 (4.74) | 30.3 (13.0) |
| 80–85 | 5.24 | 25.36 | 40.31 | 37.89 | 10.6 | 4.32 (12.24) | 31.3 (13.2) |
| P value | < 0.0001* | < 0.0001* | < 0.0001* | 0.013* | 0.54 | < 0.0001** | 0.013** |
|
Race and Ethnicity (N=3820) |
|||||||
| Hispanic | 0.96 | 4.18 | 10.96 | 43.14 | 11.97 | 1.85 (7.42) | 28.4 (12.0) |
| Non-Hispanic White | 0.95 | 6.24 | 12.66 | 34.10 | 7.67 | 1.88 (3.28) | 31.1 (12.7) |
| Non-Hispanic Black | 1.87 | 7.97 | 13.22 | 35.54 | 10.62 | 2.32 (5.64) | 30.5 (13.3) |
| Other | 0.0 | 3.21 | 9.55 | 27.66 | 6.98 | 1.34 (1.34) | 33.6 (12.7) |
| P value | 0.12 | 0.13 | 0.53 | 0.046* | 0.11 | 0.08 | < 0.0001** |
|
Education (N=3815) |
|||||||
| Less than High School | 1.49 | 8.49 | 16.96 | 35.99 | 10.98 | 2.42 (7.72) | 29.8 (12.9) |
| High School | 0.93 | 5.76 | 12.32 | 37.76 | 9.46 | 1.82 (3.13) | 30.7 (12.9) |
| More than High School | 0.88 | 5.54 | 11.61 | 33.08 | 6.94 | 1.69 (2.58) | 30.8 (12.5) |
| P value | 0.43 | 0.059 | 0.0023* | 0.11 | 0.0071* | 0.002** | 0.036** |
|
Marital Status (N=3816) |
|||||||
| Ever Married | 1.05 | 6.33 | 13.00 | 34.83 | 8.22 | 1.99 (5.06) | 30.5 (12.7) |
| Never Married | 0.59 | 3.86 | 7.40 | 33.65 | 8.28 | 1.53 (3.08) | 30.7 (13.1) |
| P value | 0.32 | 0.078 | 0.0077* | 0.73 | 0.97 | < 0.0001** | 0.98 |
|
Household income ($) (N=3628) |
|||||||
| < 20000 | 1.26 | 8.47 | 15.27 | 37.67 | 10.45 | 2.09 (4.17) | 30.1 (13.0) |
| ≥ 20000 | 0.90 | 5.73 | 11.85 | 34.25 | 7.90 | 1.87 (4.98) | 30.7 (12.6) |
| P value | 0.35 | 0.0002* | 0.038* | 0.29 | 0.037* | 0.42 | 0.08 |
|
Smoking history (N=2644) |
|||||||
| Current smoker | 0.63 | 3.46 | 10.10 | 37.52 | 9.50 | 1.56 (3.28) | 29.5 (12.1) |
| Former smoker | 0.92 | 7.33 | 14.09 | 31.50 | 7.26 | 2.00 (3.61) | 31.6 (13.1) |
| Never smoker | 0.90 | 6.63 | 12.64 | 31.94 | 7.04 | 1.92 (3.56) | 31.5 (12.8) |
| P value | 0.86 | 0.018* | 0.18 | 0.13 | 0.33 | 0.0004** | 0.004** |
|
Body mass index (kg/m2) (N=3713) |
|||||||
| < 25 | 1.48 | 7.39 | 12.65 | 36.02 | 8.69 | 2.12 (4.07) | 30.5 (12.9) |
| 25-<30 | 1.05 | 6.70 | 13.94 | 34.07 | 8.42 | 2.00 (4.21) | 30.3 (12.4) |
| 30+ | 0.48 | 3.99 | 10.03 | 34.52 | 7.71 | 1.49 (2.11) | 30.7 (12.9) |
| P value | 0.15 | 0.0013* | 0.013* | 0.73 | 0.77 | < 0.0001** | 0.09 |
%fPSA = Percent free prostate specific antigen; SD=Standard deviation; tPSA = Total prostate specific antigen;
P < 0.05 based on Chi-square tests of independence;
P < 0.05 based on ANOVA test with log-transformed tPSA and %fPSA values.
As shown in Table 4log-transformed tPSA differed significantly according to quartiles of retinol, α-carotene, β-carotene, lycopene and total carotenoids. Similarly, log-transformed percentfPSA differed according to quartiles of retinol alone. Cross-tabulation of PSA biomarkers (defined as dichotomous variables) against quartiles of retinol, retinyl esters and carotenoids suggested that two tPSA thresholds, namely tPSA > 4 ng/mL and tPSA > 2.5 ng/mL, were directly related to β-carotene and inversely related to lycopene.
Table 4.
Logistic regression and one-way ANOVA models for unadjusted associations of serum levels of retinol, retinyl esters and carotenoids (defined as quartiles) with prostate specific antigen levels – 2001–2006 National Health and Nutrition Examination Survey*
| tPSA > 10 ng/ml (OR (95% CI), %) |
tPSA > 4 ng/ml (%, OR (95% CI), %) |
tPSA > 2.5 ng/ml (OR (95% CI), %) |
%fPSA < 25% (OR 95% CI), %) |
%fPSA < 15% (OR (95% CI), %) |
tPSA Mean (SD) |
%fPSA Mean (SD) |
|
|---|---|---|---|---|---|---|---|
|
Retinol (µmol/L) (N=3810) |
|||||||
| Q1: 0.024–1.85 | 1.00, 1.3% |
1.00, 5.8% |
1.00, 12.7% |
1.00, 38.9% |
1.00, 8.7% |
1.76 (3.36) | 29.8 (12.6) |
| Q2: 1.85–2.21 | 0.42 (0.16–1.09), 0.5% |
0.88 (0.58–1.32), 5.1% |
0.79 (0.58–1.07), 10.3% |
0.75 (0.56–0.99)*, 32.3% |
0.94 (0.66–1.33), 8.2% |
1.82 (3.21) | 30.4 (12.8) |
| Q3: 2.21–2.62 | 0.98 (0.41–2.39), 1.2% |
0.99 (0.64–1.53), 5.7% |
0.88 (0.64–1.22), 11.5% |
0.79 (0.65–0.97), 33.7% |
1.05 (0.71–1.53), 9.1% |
1.96 (4.13) | 29.55 (11.71) |
| Q4: 2.62–8.76 | 0.79 (0.34–1.88), 0.9% |
1.38 (0.88–2.17), 7.8% |
1.23 (0.93–1.65), 15.3% |
0.83 (0.66–1.04), 34.6% |
0.81 (0.53–1.22), 7.2% |
2.17 (7.18) | 32.3 (13.5) |
| P value | 0.24 | 0.09 | 0.008* | 0.13 | 0.46 | 0.0014** | < 0.0001** |
|
Retinyl esters (µmol/L) (N=3573) |
|||||||
| Q1: 0.0059–0.048 | 1.00, 1.5% |
1.00, 6.3% |
1.00, 12.8% |
1.00, 30.8% |
1.00, 7.9% |
1.93 (3.74) | 30.9 (12.9) |
| Q2: 0.048–0.073 | 0.74 (0.31–1.76), 1.1% |
1.04 (0.69–1.57), 6.5% |
1.06 (0.81–1.38), 13.4% |
1.25 (0.93–1.68), 35.8% |
1.00 (0.62–1.65), 7.9% |
2.03 (4.11) | 30.6 (12.9) |
| Q3: 0.073–0.12 | 0.37 (0.11–1.21), 0.6% |
0.93 (0.59–1.43), 5.9% |
0.99 (0.74–1.37), 12.7% |
1.33 (0.99–1.77), 37.1% |
1.27 (0.77–2.11), 9.9% |
1.67 (2.21) | 30.1 (12.6) |
| Q4: 0.12–2.20 | 0.45 (0.19–1.09), 0.7% |
0.86 (0.56–1.33), 5.5% |
0.86 (0.67–1.12), 11.3% |
1.22 (0.93–1.59), 35.2% |
0.84 (0.53–1.35), 6.7% |
2.02 (7.41) | 30.6 (12.5) |
| P value | 0.14 | 0.80 | 0.56 | 0.15 | 0.33 | 0.98 | 0.47 |
|
α-carotene (µmol/L) (N=3804) |
|||||||
| Q1: 0.0039–0.026 | 1.00, 1.2% |
1.00, 5.8% |
1.00, 11.8% |
1.00, 36.2% |
1.00, 10.2% |
1.76 (3.41) | 30.3 (13.2) |
| Q2: 0.026–0.048 | 0.62 (0.24–1.60), 0.7% |
0.98 (0.67–1.42), 5.6% |
0.97 (0.72–1.30), 11.5% |
1.04 (0.78–1.38), 37.1% |
0.80 (0.54–1.19), 8.4% |
1.94 (3.98) | 30.6 (13.1) |
| Q3: 0.048–0.089 | 0.63 (0.24–1.69), 0.8% |
0.98 (0.66–1.46), 5.7% |
1.08 (0.81–1.44), 12.6% |
0.86 (0.69–1.08), 32.9% |
0.80 (0.57–1.14), 8.3% |
1.83 (3.27) | 30.8 (12.5) |
| Q4: 0.089–1.80 | 1.14 (0.45–2.93), 1.3% |
1.31 (0.89–1.92), 7.4% |
1.23 (0.89–1.68), 14.1% |
0.83 (0.60–1.14), 32.0% |
0.54 (0.36–0.80), 5.8% |
2.18 (7.24) | 30.4 (12.0) |
| P value | 0.41 | 0.32 | 0.39 | 0.37 | 0.038* | 0.0086** | 0.47 |
|
β-carotene (µmol/L) (N=3696) |
|||||||
| Q1: 0.012–0.14 | 1.00, 0.5% |
1.00, 4.9% |
1.00, 9.9% |
1.00, 36.5% |
1.00, 9.0% |
1.52 (2.11) | 30.2 (12.9) |
| Q2: 0.14–0.24 | 2.38 (0.73–7.75), 1.1% |
1.09 (0.75–1.58), 5.3% |
1.25 (0.89–1.74), 12.2% |
0.92 (0.70–1.19), 34.5% |
1.08 (0.65–1.78), 9.6% |
1.82 (3.76) | 30.8 (13.1) |
| Q3: 0.24–0.40 | 3.02 (0.82–11.1), 1.4% |
1.29 (0.86–1.94), 6.3% |
1.47 (1.10–1.95), 13.9% |
0.81 (0.65–1.02), 31.9% |
0.63 (0.43–0.93), 5.9% |
2.01 (3.38) | 31.1 (12.9) |
| Q4: 0.40–6.39 | 2.12 (0.59–7.64),1.0% | 1.69 (1.08–2.67), 8.1% |
1.53 (1.11–2.10), 14.5% |
0.91 (0.74–1.14), 34.5% |
0.94 (0.60–1.49), 8.5% |
2.37 (7.86) | 30.0 (12.1) |
| P value | 0.27 | 0.047* | 0.028* | 0.36 | 0.13 | < 0.0001** | 0.22 |
|
β-cryptoxanthin (µmol/L) M(N=3789) |
|||||||
| Q1: 0.0025–0.080 | 1.00, 0.7% |
1.00, 5.9% |
1.00, 11.7% |
1.00, 35.5% |
9.5, 1.00 |
3.86 (12.84) | 27.4 (12.9) |
| Q2: 0.080–0.13 | 1.97 (0.76–5.12), 1.3% |
1.11 (0.78–1.59), 6.5% |
1.13 (0.82–1.56), 13.1% |
0.94 (0.77–1.15), 34.1% |
0.79 (0.55–1.14), 7.7% |
1.87 (3.23) | 30.6 (13.2) |
| Q3: 0.13–0.22 | 1.87 (0.57–6.09), 1.2% |
0.97 (0.66–1.43), 5.7% |
1.15 (0.89–1.48), 13.3% |
0.95 (0.77–1.16), 34.3% |
0.83 (0.57–1.21), 8.0% |
2.19 (7.74) | 31.2 (12.7) |
| Q4: 0.22–2.73 | 1.19 (0.51–2.80), 0.8% |
1.12 (0.76–1.67), 6.5% |
1.03 (0.75–1.41), 12.0% |
0.93 (0.78–1.12), 33.9% |
0.77 (0.51–1.16), 7.5% |
1.95 (3.38) | 30.5 (12.5) |
| P value | 0.43 | 0.84 | 0.69 | 0.89 | 0.43 | 0.18 | 0.12 |
|
lutein+zeaxanthin (µmol/L) (N=3803) |
|||||||
| Q1: 0.03–0.19 | 1.00, 0.7% |
1.00, 5.9% |
1.00, 11.2% |
1.00, 33.2% |
1.00, 8.7% |
2.12 (7.72) | 30.9 (13.4) |
| Q2: 0.19–0.27 | 1.99 (0.68–5.79), 1.5% |
0.98 (0.68–1.42), 5.8% |
1.27 (0.94–1.72), 13.8% |
1.04 (0.84–1.27), 35.3% |
1.07 (0.72–1.58), 9.3% |
1.84 (3.27) | 30.6 (12.6) |
| Q3: 0.27–0.37 | 1.54 (0.71–3.32), 1.1% |
1.06 (0.74–1.52), 6.2% |
1.19 (0.89–1.58), 13.1% |
1.01 (0.76–1.32), 34.2% |
0.88 (0.56–1.39), 7.8% |
1.94 (3.13) | 29.97 (12.60) |
| Q4: 0.37–1.95 | 0.84 (0.33–2.10), 0.6% |
1.15 (0.81–1.63), 6.7% |
1.07 (0.81–1.43), 11.9% |
0.87 (0.71–1.06), 35.0% |
0.76 (0.52–1.12), 6.8% |
1.81 (3.12) | 30.5 (12.3) |
| P value | 0.19 | 0.81 | 0.30 | 0.88 | 0.38 | 0.19 | 0.57 |
|
Lycopene (µmol/L) (N=3803) |
|||||||
| Q1: 0.0067–0.23 | 1.00, 1.2% |
1.00, 8.3% |
1.00, 15.3% |
1.00, 33.2% |
1.00, 7.7% |
2.38 (7.59) | 30.8 (12.9) |
| Q2: 0.23–0.35 | 0.76 (0.37–1.54), 0.9% |
0.85 (0.61–1.19), 7.2% |
0.89 (0.66–1.22), 13.9% |
1.09 (0.88–1.36), 35.3% |
1.09 (0.66–1.79), 8.3% |
1.90 (3.33) | 31.2 (13.2) |
| Q3: 0.35–0.52 | 0.34 (0.12–0.99), 0.9% |
0.59 (0.39–0.89), 5.1% |
0.75 (0.52–1.07), 11.9% |
1.05 (0.81–1.35), 34.3% |
1.19 (0.73–1.95), 9.0% |
1.74 (3.36) | 29.7 (12.4) |
| Q4: 0.52–1.49 | 0.44 (0.16–1.19), 1.1% |
0.57 (0.39–0.83), 4.9% |
0.64 (0.46–0.88), 10.3% |
1.08 (0.83–1.40), 35.0% |
1.02 (0.60–1.71), 7.8% |
1.69 (3.18) | 30.3 (12.3) |
| P value | 0.85 | 0.007* | 0.033* | 0.33 | 0.84 | 0.0023** | 0.14 |
|
Total carotenoids (µmol/L) (N=3685) |
|||||||
| Q1: 0.082–0.81 | 1.00, 1.2% |
1.00, 6.5% |
1.00, 12.6% |
1.00, 35.0% |
1.00, 9.1% |
1.94 (3.69) | 30.8 (13.3) |
| Q2: 0.81–1.14 | 0.72 (0.26–2.04), 0.9% |
0.94 (0.62–1.43), 6.2% |
1.06 (0.78–1.44), 13.2% |
1.06 (0.83–1.36), 36.5% |
1.09 (0.66–1.79), 9.9% |
1.75 (2.74) | 30.5 (12.8) |
| Q3: 1.14–1.58 | 0.71 (0.25–1.99), 0.9% |
0.70 (0.49–1.00), 4.7% |
0.87 (0.64–1.17), 11.1% |
0.86 (0.64–1.14), 31.6% |
0.69 (0.43–1.13), 6.5% |
1.78 (3.47) | 31.2 (12.8) |
| Q4: 1.58–6.92 | 0.93 (0.41–2.10), 1.1% |
1.14 (0.77–1.68), 7.3% |
1.11 (0.82–1.50), 13.8% |
0.98 (0.81–1.20), 34.7% |
0.85 (0.52–1.39), 7.9% |
2.25 (7.68) | 29.7 (12.1) |
| P value | 0.84 | 0.13 | 0.41 | 0.33 | 0.25 | 0.035** | 0.14 |
%fPSA = Percent free prostate specific antigen; SEM=Standard error of the mean; tPSA = Total prostate specific antigen;
P < 0.05 based on Chi-square tests of independence;
P < 0.05 based on one-way ANOVA tests with log-transformed tPSA and %fPSA values.
Table 5 displays regression models for fully-adjusted associations of serum levels of retinol, retinyl esters and carotenoids with PSA levels. Multivariate logistic regression models suggested that high serum levels of retinyl esters (tPSA>10 ng/ml: Q4vs.Q1→OR=0.38, 95% CI: 0.14, 1.00) and α-carotene (percentfPSA<15 percent: Q4vs.Q1→OR=0.49, 95% CI: 0.32, 0.76) were associated with a lower odds of PCa detection. Whereas crude analyses implied an inverse relationship between high serum level of lycopene and PCa detection (tPSA>2.5 ng/ml: Q4vs.Q1→OR=0.64, 95% CI: 0.46, 0.88), the same exposure was found to be directly related to screen-detected PCa after adjustment for a priori confounders (tPSA>2.5 ng/ml: Q4vs.Q1→OR=1.49, 95% CI: 1.01, 2.14). Test-for-trend analyses in fully adjusted logistic regression models suggest that tPSA > 10 ng/mL was inversely associated with increasing levels of lutein+zeaxanthin (Ptrend=0.04) and that %fPSA < 15 ng/ml was inversely associated with increasing levels of α-carotene (Ptrend=0.004). Retinol and other carotenoids were not significantly associated with PSA biomarkers defined as dichotomous outcomes.
Table 5.
Regression models for fully-adjusted associations of serum levels of retinol, retinyl esters and carotenoids (defined as quartiles) with prostate specific antigen levels – 2001–2006 National Health and Nutrition Examination Survey
| tPSA > 10 ng/ml | tPSA > 4 ng/ml | tPSA > 2.5 ng/ml | %fPSA < 25 ng/ml | %fPSA < 15 ng/ml | |
|---|---|---|---|---|---|
| Retinol (µmol/L) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) |
| N=3416 | N=3509 | N=3509 | N=3509 | N=3509 | |
| Q1: 0.024–1.85 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Q2: 1.85–2.21 | 0.41 (0.14,1.23) | 0.80 (0.50,1.28) | 0.69 (0.48,0.97) | 0.77 (0.58,1.03) | 0.95 (0.64,1.40) |
| Q3: 2.21–2.62 | 0.76 (0.25,2.31) | 0.79 (0.48,1.31) | 0.75 (0.52,1.11) | 0.83 (0.68,1.00) | 1.08 (0.71,1.66) |
| Q4: 2.62–8.76 | 0.63 (0.22,1.79) | 1.17 (0.68,2.01) | 1.02 (0.73,1.44) | 0.88 (0.69,1.13) | 0.93 (0.58,1.49) |
| Ptrend | 0.66 | 0.44 | 0.47 | 0.57 | 0.92 |
| Retinyl esters (µmol/L) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) |
| N=3202 | N=3291 | N=3291 | N=3291 | N=3291 | |
| Q1: 0.0059–0.048 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Q2: 0.048–0.073 | 0.42 (0.13,1.30) | 0.90 (0.53,1.54) | 0.96 (0.67,1.38) | 1.24 (0.91,1.70) | 0.93 (0.54,1.58) |
| Q3: 0.073–0.12 | 0.37 (0.09,1.38) | 0.92 (0.60,1.42) | 1.02 (0.76,1.37) | 1.33 (0.99,1.80) | 1.25 (0.74,2.10) |
| Q4: 0.12–2.20 | 0.38 (0.14,1.00) | 0.84 (0.53,1.33) | 0.82 (0.60,1.12) | 1.24 (0.92,1.66) | 0.93 (0.56,1.53) |
| Ptrend | 0.069 | 0.48 | 0.30 | 0.11 | 0.93 |
| α-carotene (µmol/L) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) |
| N=3411 | N=3504 | N=3504 | N=3504 | N=3504 | |
| Q1: 0.0039–0.026 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Q2: 0.026–0.048 | 0.52 (0.18,1.55) | 0.88 (0.58,1.34) | 0.88 (0.65,1.21) | 1.06 (0.79,1.41) | 0.83 (0.55,1.25) |
| Q3: 0.048–0.089 | 0.47 (0.15,1.44) | 0.78 (0.52,1.17) | 0.89 (0.66,1.20) | 0.83 (0.65,1.05) | 0.79 (0.53,1.19) |
| Q4: 0.089–1.80 | 0.50 (0.16,1.61) | 0.96 (0.62,1.49) | 0.95 (0.67,1.35) | 0.78 (0.56,1.12) | 0.49 (0.32,0.76) |
| Ptrend | 0.26 | 0.80 | 0.82 | 0.068 | 0.004 |
| β-carotene (µmol/L) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) |
| N=3308 | N=3399 | N=3399 | N=3399 | N=3399 | |
| Q1: 0.012–0.14 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Q2: 0.14–0.24 | 1.98 (0.57,6.93) | 0.92 (0.59,1.43) | 1.04 (0.72,1.51) | 0.93 (0.72,1.20) | 1.17 (0.70,1.95) |
| Q3: 0.24–0.40 | 1.94 (0.48,7.88) | 0.92 (0.58,1.46) | 1.11 (0.77,1.59) | 0.83 (0.64,1.07) | 0.67 (0.44,1.02) |
| Q4: 0.40–6.39 | 0.67 (0.17,2.61) | 0.95 (0.57,1.60) | 0.98 (0.67,1.42) | 0.87 (0.68,1.12) | 0.95 (0.58,1.54) |
| Ptrend | 0.42 | 0.88 | 0.96 | 0.17 | 0.36 |
| β-cryptoxanthin (µmol/L) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) |
| N=3399 | N=3492 | N=3492 | N=3492 | N=3492 | |
| Q1: 0.0025–0.080 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Q2: 0.080–0.13 | 1.85 (0.66,5.16) | 1.04 (0.68,1.59) | 1.12 (0.78,1.59) | 0.98 (0.79,1.20) | 0.83 (0.55,1.25) |
| Q3: 0.13–0.22 | 1.41 (0.35,5.67) | 0.87 (0.56,1.36) | 1.09 (0.81,1.49) | 0.99 (0.79,1.25) | 0.83 (0.53,1.28) |
| Q4: 0.22–2.73 | 0.89 (0.30,2.61) | 1.14 (0.73,1.78) | 1.04 (0.72,1.50) | 0.93 (0.75,1.14) | 0.76 (0.46,1.23) |
| Ptrend | 0.76 | 0.78 | 0.82 | 0.59 | 0.28 |
| lutein+zeaxanthin (µmol/L) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) |
| N=3411 | N=3504 | N=3504 | N=3504 | N=3504 | |
| Q1: 0.03–0.19 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Q2: 0.19–0.27 | 1.87 (0.59,5.91) | 1.08 (0.71,1.64) | 1.38 (0.99,1.92) | 1.08 (0.87,1.34) | 1.17 (0.79,1.74) |
| Q3: 0.27–0.37 | 0.86 (0.34,2.17) | 1.05 (0.70,1.56) | 1.17 (0.83,1.65) | 0.98 (0.74,1.31) | 0.92 (0.58,1.44) |
| Q4: 0.37–1.95 | 0.52 (0.18,1.52) | 1.16 (0.78,1.71) | 1.03 (0.74,1.45) | 0.83 (0.69,1.01) | 0.78 (0.53,1.18) |
| Ptrend | 0.040 | 0.51 | 0.92 | 0.064 | 0.17 |
| Lycopene (µmol/L) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) |
| N=3411 | N=3504 | N=3504 | N=3504 | N=3504 | |
| Q1: 0.0067–0.23 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Q2: 0.23–0.35 | 0.78 (0.36,1.70) | 1.15 (0.82,1.61) | 1.25 (0.94,1.68) | 1.17 (0.94,1.44) | 1.15 (0.69,1.90) |
| Q3: 0.35–0.52 | 0.28 (0.09,0.84) | 0.95 (0.65,1.39) | 1.26 (0.86,1.85) | 1.16 (0.89,1.51) | 1.28 (0.75,2.17) |
| Q4: 0.52–1.49 | 0.99 (0.35,2.76) | 1.39 (0.92,2.09) | 1.49 (1.04,2.14) | 1.22 (0.92,1.61) | 1.18 (0.67,2.06) |
| Ptrend | 0.67 | 0.23 | 0.057 | 0.22 | 0.54 |
| Total carotenoids (µmol/L) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) |
| N=3299 | N=3390 | N=3390 | N=3390 | N=3390 | |
| Q1: 0.082–0.81 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Q2: 0.81–1.14 | 0.68 (0.21,2.14) | 0.89 (0.54,1.49) | 1.11 (0.82,1.49) | 1.14 (0.89,1.47) | 1.06 (0.61,1.85) |
| Q3: 1.14–1.58 | 0.75 (0.23,2.46) | 0.79 (0.57,1.13) | 1.00 (0.78,1.28) | 0.88 (0.65,1.19) | 0.72 (0.42,1.21) |
| Q4: 1.58–6.92 | 0.54 (0.20,1.42) | 1.10 (0.70,1.73) | 1.17 (0.82,1.68) | 1.00 (0.80,1.26) | 0.87 (0.51,1.47) |
| Ptrend | 0.31 | 0.75 | 0.50 | 0.48 | 0.29 |
CI=confidence interval; Q1=1st quartile; Q2=2nd quartile; Q3=3rd quartile; Q4=4th quartile; OR=odds ratio; tPSA = Total prostate specific antigen; %fPSA = Percent free prostate specific antigen;
Multivariate regression models were adjusted for age (‘40–49’, ‘50–59’, ‘60–69’, ‘70–79’, ‘80+’ years), race/ethnicity (‘Hispanic’, ‘Non-Hispanic White’, ‘Non-Hispanic Black, Other’), education (‘Less than High School’, ‘High School’, ‘More than High School’), marital status (‘Ever married’, ‘Never married’), household income (‘< $20,000’, ‘≥ $20000’) and body mass index (‘<25 kg/m2’, ‘25-<30 kg/m2’, ‘≥ 30 kg/m2’).
Finally, multiple linear regression models were constructed whereby outcomes were defined as continuous variables and exposures were defined as quartiles or continuous variables whereby fractional polynomials and quadratic terms were explored (continuous outcomes). After adjustment for known demographic, socioeconomic and lifestyle confounders, no consistent trends could be identified through linear regression models for log-transformed tPSA and percentfPSA. When added to multivariable linear regression models, most quadratic terms were found to be statistically non-significant. One exception was the model where log-transformed percentfPSA was the dependent variable and beta-carotene was the independent variable. This finding may suggest that the relationship between β-carotene and percentfPSA is non-linear (data not shown).
DISCUSSION
In this cross-sectional study of U.S. men, we evaluated whether retinol, retinyl esters, α-carotene, β-carotene, β-cryptoxanthin, lutein, zeaxanthin, lycopene and total carotenoids were associated with PCa detection, using established threshold values for tPSA and percentfPSA. Multivariate logistic models were constructed after adjustment for age, race/ethnicity, education, marital status, household income and BMI. Three key results were obtained. First, the odds of having tPSA > 10 ng/ml were significantly reduced in the upper versus the lower quartiles of retinyl esters. Second, the odds of having percentfPSA<15 percent were significantly reduced in the upper versus the lower quartiles of α-carotene. Third, the odds of having tPSA>2.5 ng/ml were significantly increased in the upper versus the lower quartiles of lycopene.
The findings that serum levels of retinyl esters and α-carotene were inversely related to serum tPSA and percentfPSA threshold levels, respectively, are consistent with the idea that carotenoids have anti-oxidative properties that can reverse DNA damage produced by inflammatory processes leading to screen-detected PCa (29). This protective effect is unlikely due to physiological factors such as increased plasma volume or reduced testosterone levels, since BMI was taken into account in the multivariate analysis (2, 23).
Carotenoids are naturally occurring tetraterpenoid (40 carbon atoms) organic pigments found mostly in photosynthetic organisms, namely plants as well as some algae, fungi and bacteria (30). To date, over 600 carotenoids have been identified and classified as xanthophils (lutein, zeaxanthin) which contain oxygen or carotenes (α-carotene, β-carotene, lycopene) which are devoid of oxygen (30). In humans, carotenoids can enhance the immune system and have anti-oxidative properties as they are efficient free-radical scavengers (30).
The finding that lycopene was positively associated with a tPSA threshold level is inconsistent with current knowledge (31–42). Previous studies have mostly implied either no association or a protective effect of lycopene against PCa. For instance, a recent meta-analysis of seven studies suggested that serum lycopene was inversely related to PCa risk [RR=0.74 (0.59–0.92)] (43). In our study, it is plausible that the distinct relationships observed between serum levels of lycopene and tPSA before and after adjustment for selected characteristics is due to strong confounding by at least one of the variables that we adjusted for. We performed separate analyses for each of the selected characteristics and identified age as being the strongest confounder, based on changes in the estimated odds ratios for the hypothesized relationship. In addition, adjusting for age, but not for other characteristics, was sufficient to change the direction of the relationship between lycopene and tPSA from inverse in the unadjusted model to direct in the adjusted model. In our study, unadjusted results overwhelmingly support an inverse association between lycopene and tPSA level, as hypothesized. However, this association changed to positive after simply adjusting for age. Confounding bias by older age is plausible as it is associated with both higher PSA levels and potentially higher lycopene intake (through supplements or diet) as a consequence of prostate health awareness. These relationships would result in an upward bias, as is seen in the adjusted models. Major epidemiologic studies of the inverse association between lycopene intake and PCa were published in the late-1990's and early-2000's, which could have influenced lycopene intake among older men with higher PSA levels (and perhaps undiagnosed, underlying disease) in the current study period from 2001–2006. Furthermore, the effect of lycopene at the 2.5 ng/mL threshold was shown to be different from its effect at the 4 ng/mL and 10 ng/mL thresholds of tPSA. This finding implies the need for further investigation into the possibility of a non-linear relationship between lycopene and tPSA levels. According to Erdman and colleagues (44), lycopene has been shown to be the most potent in vitro antioxidant of all carotenoids tested; yet, in vivo studies do not support the hypothesis that lycopene (found at low concentrations in animal and human tissues) can protect against PCa through its anti-oxidative properties (44). The authors suggest alternative mediators of the protective effect of lycopene on screen-detected PCa, namely apoptosis, cell cycle inhibition, insulin-like growth factor 1 axis, gap junction communication, androgen status, detoxification enzymes and C-reactive protein (44).
The protective role of various carotenoids against chronic disease development remains controversial. Consumption of antioxidants, which include various carotenoids, through supplementation is not likely to be beneficial and excessive use of β-carotene may even be harmful (30). For instance, supplementation studies that used mega-doses of β-carotene have shown an increased risk of lung cancer among smokers (45–47). By contrast, the risk of dying from chronic diseases may be reduced among individuals who consume diets rich in carotenoids from natural sources, such as fruits and vegetables (30). Epidemiologic studies have shown that individuals who consume high levels of β-carotene in their diet and those with high levels of β-carotene in their blood have significantly reduced risk of lung cancer (30, 48–51). Because foods high in carotenoids are generally low in fat, gastrointestinal absorption and bioavailability of carotenoids may be enhanced by consumption of Gac fruit, crude palm oil, avocado fruit or oil (30).
Our study results should be interpreted with caution and in light of several limitations. First of all, we performed secondary analyses of the 2001–2006 NHANES, which is a series of cross-sectional studies. Thus, the temporal relationship between the main exposures (retinol, retinyl esters, carotenoids) and outcomes (tPSA and percentfPSA biomarkers) could not be ascertained through a cross-sectional design. Second, the 2001–2006 waves of NHANES did not include information on diagnosis and outcome of PCa, but only PSA test results. Third, a high percentage of NHANES men 40 years and older did not participate in PSA testing. Fourth, serum carotenoid levels may not correlate with actual levels of consumption through diet or supplementation. Finally, the statistically significant associations that were observed between selected carotenoids and PSA biomarkers cannot be ruled out as chance findings, given the large number of hypotheses being tested and the fact that two out of the three significant results were observed where the outcome was very sparse (tPSA>10 ng/mL and percentfPSA<15 percent).
In conclusion, serum levels of α-carotene, retinyl esters and lycopene were shown to be significantly associated with PSA biomarkers in the general population of U.S. men. Monitoring specific antioxidant levels may be helpful in early detection of PCa. Further studies are needed to elucidate the specific roles of carotenoids from various dietary sources for the purpose of PCa prevention.
Acknowledgments
ACKNOWLEDGMENTS AND FINANCIAL SUPPORT
This research was partly supported by the Intramural Research Program of the NIH, National Institute on Aging. We would like to thank Dr. Larry Brant and Dr. Joshua Goh for providing useful comments regarding the content of the manuscript.
Abbreviations
- BMI
body mass index
- CI
confidence interval
- DRE
Digital rectal exam
- fPSA
Free prostate-specific antigen
- %fPSA
Percent free prostate-specific antigen
- Q1
1st quartile
- Q2
2nd quartile
- Q3
3rd quartile
- Q4
4th quartile
- NHANES
National Health and Nutrition Examination Survey
- OR
odds ratio
- PCa
Prostate cancer
- PSA
Prostate-specific antigen
- ROS
Reactive oxygen species
- tPSA
Total prostate-specific antigen.
REFERENCES
- 1.Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA Cancer J Clin. Sep-Oct;60(5):277–300. doi: 10.3322/caac.20073. [DOI] [PubMed] [Google Scholar]
- 2.Culp S, Porter M. The effect of obesity and lower serum prostate-specific antigen levels on prostate-cancer screening results in American men. BJU Int. 2009 Nov;104(10):1457–1461. doi: 10.1111/j.1464-410X.2009.08646.x. [DOI] [PubMed] [Google Scholar]
- 3.Hegarty NJ, Fitzpatrick JM, Richie JP, Scardino PT, deVere White RW, Schroder FH, et al. Future prospects in prostate cancer. Prostate. 1999 Sep 1;40(4):261–268. doi: 10.1002/(sici)1097-0045(19990901)40:4<261::aid-pros8>3.0.co;2-h. [DOI] [PubMed] [Google Scholar]
- 4.Screening for prostate cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2008 Aug 5;149(3):185–191. doi: 10.7326/0003-4819-149-3-200808050-00008. [DOI] [PubMed] [Google Scholar]
- 5.Lacher DA, Thompson TD, Hughes JP, Saraiya M. Total, free, and percent free prostate-specific antigen levels among U.S. men, 2001-04. Adv Data. 2006 Dec;4(379):1–12. [PubMed] [Google Scholar]
- 6.Saraiya M, Kottiri BJ, Leadbetter S, Blackman D, Thompson T, McKenna MT, et al. Total and percent free prostate-specific antigen levels among U.S. men, 2001–2002. Cancer Epidemiol Biomarkers Prev. 2005 Sep;14(9):2178–2182. doi: 10.1158/1055-9965.EPI-05-0206. [DOI] [PubMed] [Google Scholar]
- 7.Singer EA, Palapattu GS, van Wijngaarden E. Prostate-specific antigen levels in relation to consumption of nonsteroidal anti-inflammatory drugs and acetaminophen: results from the 2001–2002 National Health and Nutrition Examination Survey. Cancer. 2008 Oct 15;113(8):2053–2057. doi: 10.1002/cncr.23806. [DOI] [PubMed] [Google Scholar]
- 8.Bostwick DG, Burke HB, Djakiew D, Euling S, Ho SM, Landolph J, et al. Human prostate cancer risk factors. Cancer. 2004 Nov 15;101(10 Suppl):2371–2490. doi: 10.1002/cncr.20408. [DOI] [PubMed] [Google Scholar]
- 9.Boyle P, Severi G, Giles GG. The epidemiology of prostate cancer. Urol Clin North Am. 2003 May;30(2):209–217. doi: 10.1016/s0094-0143(02)00181-7. [DOI] [PubMed] [Google Scholar]
- 10.Gallagher RP, Fleshner N. Prostate cancer, 3. Individual risk factors. CMAJ. 1998 Oct 6;159(7):807–813. [PMC free article] [PubMed] [Google Scholar]
- 11.Gronberg H. Prostate cancer epidemiology. Lancet. 2003 Mar 8;361(9360):859–864. doi: 10.1016/S0140-6736(03)12713-4. [DOI] [PubMed] [Google Scholar]
- 12.Hsing AW, Devesa SS. Trends and patterns of prostate cancer: what do they suggest? Epidemiol Rev. 2001;23(1):3–13. doi: 10.1093/oxfordjournals.epirev.a000792. [DOI] [PubMed] [Google Scholar]
- 13.Levy IG, Iscoe NA, Klotz LH. Prostate cancer: 1. The descriptive epidemiology in Canada. CMAJ. 1998 Sep 8;159(5):509–513. [PMC free article] [PubMed] [Google Scholar]
- 14.Routh JC, Leibovich BC. Adenocarcinoma of the prostate: epidemiological trends, screening, diagnosis, and surgical management of localized disease. Mayo Clin Proc. 2005 Jul;80(7):899–907. doi: 10.4065/80.7.899. [DOI] [PubMed] [Google Scholar]
- 15.Stotts RC. Cancers of the prostate, penis, and testicles: epidemiology, prevention, and treatment. Nurs Clin North Am. 2004 Jun;39(2):327–340. doi: 10.1016/j.cnur.2004.03.002. [DOI] [PubMed] [Google Scholar]
- 16.Hsing AW, Chokkalingam AP. Prostate cancer epidemiology. Front Biosci. 2006;11:1388–1413. doi: 10.2741/1891. [DOI] [PubMed] [Google Scholar]
- 17.Nelson WG, De Marzo AM, Isaacs WB. Prostate cancer. N Engl J Med. 2003 Jul 24;349(4):366–381. doi: 10.1056/NEJMra021562. [DOI] [PubMed] [Google Scholar]
- 18.De Marzo AM, DeWeese TL, Platz EA, Meeker AK, Nakayama M, Epstein JI, et al. Pathological and molecular mechanisms of prostate carcinogenesis: implications for diagnosis, detection, prevention, and treatment. J Cell Biochem. 2004 Feb 15;91(3):459–477. doi: 10.1002/jcb.10747. [DOI] [PubMed] [Google Scholar]
- 19.Platz EA, De Marzo AM. Epidemiology of inflammation and prostate cancer. J Urol. 2004 Feb;171(2 Pt 2):S36–S40. doi: 10.1097/01.ju.0000108131.43160.77. [DOI] [PubMed] [Google Scholar]
- 20.De Marzo AM, Platz EA, Sutcliffe S, Xu J, Gronberg H, Drake CG, et al. Inflammation in prostate carcinogenesis. Nat Rev Cancer. 2007 Apr;7(4):256–269. doi: 10.1038/nrc2090. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Adami H-O, Hunter D, Trichopoulos D. Textbook of Cancer Epidemiology. 2nd ed. New York: Oxford University Press; 2008. [Google Scholar]
- 22.Parekh N, Lin Y, Dipaola RS, Marcella S, Lu-Yao G. Obesity and prostate cancer detection: insights from three national surveys. Am J Med. Sep;123(9):829–835. doi: 10.1016/j.amjmed.2010.05.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Parekh N, Lin Y, Marcella S, Kant AK, Lu-Yao G. Associations of lifestyle and physiologic factors with prostate-specific antigen concentrations: evidence from the National Health and Nutrition Examination Survey (2001–2004) Cancer Epidemiol Biomarkers Prev. 2008 Sep;17(9):2467–2472. doi: 10.1158/1055-9965.EPI-08-0059. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Joseph DA, Thompson T, Saraiya M, Werny DM. Association Between Glomerular Filtration Rate, Free, Total, and Percent Free Prostate-specific Antigen. Urology. 2009 Sep 23; doi: 10.1016/j.urology.2009.05.100. [DOI] [PubMed] [Google Scholar]
- 25.Mondul AM, Selvin E, De Marzo AM, Freedland SJ, Platz EA. Statin drugs, serum cholesterol, and prostate-specific antigen in the National Health and Nutrition Examination Survey 2001–2004. Cancer Causes Control. May;21(5):671–678. doi: 10.1007/s10552-009-9494-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Chang SL, Harshman LC, Presti JC., Jr Impact of common medications on serum total prostate-specific antigen levels: analysis of the National Health and Nutrition Examination Survey. J Clin Oncol. Sep 1;28(25):3951–3957. doi: 10.1200/JCO.2009.27.9406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Werny DM, Saraiya M, Gregg EW. Prostate-specific antigen values in diabetic and nondiabetic US men, 2001–2002. Am J Epidemiol. 2006 Nov 15;164(10):978–83. doi: 10.1093/aje/kwj311. [DOI] [PubMed] [Google Scholar]
- 28.Royston P, Sauerbrei W. Multivariable model-building: a pragmatic approach to regression analysis based on fractional polynomials for modelling continuous variables. Hoboken, NJ: John Wiley & Sons; 2009. [Google Scholar]
- 29.Wagenlehner FM, Elkahwaji JE, Algaba F, Bjerklund-Johansen T, Naber KG, Hartung R, et al. The role of inflammation and infection in the pathogenesis of prostate carcinoma. BJU Int. 2007 Oct;100(4):733–737. doi: 10.1111/j.1464-410X.2007.07091.x. [DOI] [PubMed] [Google Scholar]
- 30.Carotenoid. [cited]; Available from: http://en.wikipedia.org/wiki/Carotenoid.
- 31.Basu A, Imrhan V. Tomatoes versus lycopene in oxidative stress and carcinogenesis: conclusions from clinical trials. Eur J Clin Nutr. 2007 Mar;61(3):295–303. doi: 10.1038/sj.ejcn.1602510. [DOI] [PubMed] [Google Scholar]
- 32.Edinger MS, Koff WJ. Effect of the consumption of tomato paste on plasma prostate-specific antigen levels in patients with benign prostate hyperplasia. Braz J Med Biol Res. 2006 Aug;39(8):1115–1119. doi: 10.1590/s0100-879x2006000800014. [DOI] [PubMed] [Google Scholar]
- 33.Haseen F, Cantwell MM, O’Sullivan JM, Murray LJ. Is there a benefit from lycopene supplementation in men with prostate cancer? A systematic review. Prostate Cancer Prostatic Dis. 2009;12(4):325–332. doi: 10.1038/pcan.2009.38. [DOI] [PubMed] [Google Scholar]
- 34.Kucuk O, Sarkar FH, Djuric Z, Sakr W, Pollak MN, Khachik F, et al. Effects of lycopene supplementation in patients with localized prostate cancer. Exp Biol Med (Maywood) 2002 Nov;227(10):881–885. doi: 10.1177/153537020222701007. [DOI] [PubMed] [Google Scholar]
- 35.Kucuk O, Sarkar FH, Sakr W, Djuric Z, Pollak MN, Khachik F, et al. Phase II randomized clinical trial of lycopene supplementation before radical prostatectomy. Cancer Epidemiol Biomarkers Prev. 2001 Aug;10(8):861–868. [PubMed] [Google Scholar]
- 36.Liu A, Pajkovic N, Pang Y, Zhu D, Calamini B, Mesecar AL, et al. Absorption and subcellular localization of lycopene in human prostate cancer cells. Mol Cancer Ther. 2006 Nov;5(11):2879–2885. doi: 10.1158/1535-7163.MCT-06-0373. [DOI] [PubMed] [Google Scholar]
- 37.Rao AV, Fleshner N, Agarwal S. Serum and tissue lycopene and biomarkers of oxidation in prostate cancer patients: a case-control study. Nutr Cancer. 1999;33(2):159–164. doi: 10.1207/S15327914NC330207. [DOI] [PubMed] [Google Scholar]
- 38.Schroder FH, Roobol MJ, Boeve ER, de Mutsert R, Zuijdgeest-van Leeuwen SD, Kersten I, et al. Randomized, double-blind, placebo-controlled crossover study in men with prostate cancer and rising PSA: effectiveness of a dietary supplement. Eur Urol. 2005 Dec;48(6):922–930. doi: 10.1016/j.eururo.2005.08.005. discussion 30-1. [DOI] [PubMed] [Google Scholar]
- 39.Schwarz S, Obermuller-Jevic UC, Hellmis E, Koch W, Jacobi G, Biesalski HK. Lycopene inhibits disease progression in patients with benign prostate hyperplasia. J Nutr. 2008 Jan;138(1):49–53. doi: 10.1093/jn/138.1.49. [DOI] [PubMed] [Google Scholar]
- 40.Stacewicz-Sapuntzakis M, Bowen PE. Role of lycopene and tomato products in prostate health. Biochim Biophys Acta. 2005 May 30;1740(2):202–205. doi: 10.1016/j.bbadis.2005.02.004. [DOI] [PubMed] [Google Scholar]
- 41.Talvas J, Caris-Veyrat C, Guy L, Rambeau M, Lyan B, Minet-Quinard R, et al. Differential effects of lycopene consumed in tomato paste and lycopene in the form of a purified extract on target genes of cancer prostatic cells. Am J Clin Nutr. Jun;91(6):1716–1724. doi: 10.3945/ajcn.2009.28666. [DOI] [PubMed] [Google Scholar]
- 42.Vaishampayan U, Hussain M, Banerjee M, Seren S, Sarkar FH, Fontana J, et al. Lycopene and soy isoflavones in the treatment of prostate cancer. Nutr Cancer. 2007;59(1):1–7. doi: 10.1080/01635580701413934. [DOI] [PubMed] [Google Scholar]
- 43.Etminan M, Takkouche B, Caamano-Isorna F. The role of tomato products and lycopene in the prevention of prostate cancer: a meta-analysis of observational studies. Cancer Epidemiol Biomarkers Prev. 2004 Mar;13(3):340–345. [PubMed] [Google Scholar]
- 44.Erdman JW, Jr, Ford NA, Lindshield BL. Are the health attributes of lycopene related to its antioxidant function? Arch Biochem Biophys. 2009 Mar 15;483(2):229–235. doi: 10.1016/j.abb.2008.10.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Lung cancer associated with beta-carotene supplementation in smokers. Prescrire Int. Jun;19(107):121. [PubMed] [Google Scholar]
- 46.Kim Y, Chongviriyaphan N, Liu C, Russell RM, Wang XD. Combined antioxidant (beta-carotene, alpha-tocopherol and ascorbic acid) supplementation increases the levels of lung retinoic acid and inhibits the activation of mitogen-activated protein kinase in the ferret lung cancer model. Carcinogenesis. 2006 Jul;27(7):1410–1419. doi: 10.1093/carcin/bgi340. [DOI] [PubMed] [Google Scholar]
- 47.Kim Y, Lian F, Yeum KJ, Chongviriyaphan N, Choi SW, Russell RM, et al. The effects of combined antioxidant (beta-carotene, alpha-tocopherol and ascorbic acid) supplementation on antioxidant capacity, DNA single-strand breaks and levels of insulinlike growth factor-1/IGF-binding protein 3 in the ferret model of lung cancer. Int J Cancer. 2007 May 1;120(9):1847–1854. doi: 10.1002/ijc.22320. [DOI] [PubMed] [Google Scholar]
- 48.Greenwald P. Beta-carotene and lung cancer: a lesson for future chemoprevention investigations? J Natl Cancer Inst. 2003 Jan 1;95(1):E1. doi: 10.1093/jnci/95.1.e1. [DOI] [PubMed] [Google Scholar]
- 49.Holick CN, Michaud DS, Stolzenberg-Solomon R, Mayne ST, Pietinen P, Taylor PR, et al. Dietary carotenoids, serum beta-carotene, and retinol and risk of lung cancer in the alpha-tocopherol, beta-carotene cohort study. Am J Epidemiol. 2002 Sep 15;156(6):536–547. doi: 10.1093/aje/kwf072. [DOI] [PubMed] [Google Scholar]
- 50.Neuhouser ML, Patterson RE, Thornquist MD, Omenn GS, King IB, Goodman GE. Fruits and vegetables are associated with lower lung cancer risk only in the placebo arm of the beta-carotene and retinol efficacy trial (CARET) Cancer Epidemiol Biomarkers Prev. 2003 Apr;12(4):350–358. [PubMed] [Google Scholar]
- 51.Satia JA, Littman A, Slatore CG, Galanko JA, White E. Long-term use of beta-carotene, retinol, lycopene, and lutein supplements and lung cancer risk: results from the VITamins And Lifestyle (VITAL) study. Am J Epidemiol. 2009 Apr 1;169(7):815–828. doi: 10.1093/aje/kwn409. [DOI] [PMC free article] [PubMed] [Google Scholar]