Skip to main content
HHS Author Manuscripts logoLink to HHS Author Manuscripts
. Author manuscript; available in PMC: 2015 Jun 24.
Published in final edited form as: J Urol. 2012 Feb 14;187(4):1228–1233. doi: 10.1016/j.juro.2011.11.086

Association Between Smoking Status, and Free, Total and Percent Free Prostate Specific Antigen

Jun Li 1,*, Trevor Thompson 1, Djenaba A Joseph 1, Viraj A Master 1
PMCID: PMC4479165  NIHMSID: NIHMS701622  PMID: 22335864

Abstract

Purpose

There are scant data available on the relationship between smoking and total prostate specific antigen, free prostate specific antigen and percent-free prostate specific antigen. Given the high prevalence of smoking and the frequency of prostate specific antigen screening, it is important to determine any association between smoking and prostate specific antigen values using nationally representative data.

Materials and Methods

Included in the final study population were 3,820 men 40 years old or older who participated in the 2001–2006 NHANES (National Health and Nutrition Examination Survey) and met the eligibility criteria for prostate specific antigen testing. The distributions of total, free and percent free prostate specific antigen were estimated by sociodemographic and clinical characteristics. Multivariate linear regression models were fit to determine the adjusted relationship between smoking and total and percent free prostate specific antigen while simultaneously controlling for these characteristics.

Results

For all ages combined the median total and free prostate specific antigen levels were 0.90 (0.81– 0.90) and 0.26 (0.25– 0.28) ng/ml, respectively. Multivariate linear regression analysis showed that total prostate specific antigen was 7.9% and 12.2% lower among current and former smokers, respectively, than among never smokers. High body mass index and diabetes were also statistically significantly associated with a lower total prostate specific antigen. Approximately a third of the men had a percent free prostate specific antigen less than 25%. Current smokers had a significantly lower percent free prostate specific antigen than former smokers.

Conclusions

Our finding that smoking is inversely associated with total prostate specific antigen may have potential implications for the interpretation of prostate specific antigen levels in men who are current or former smokers. Given the high prevalence of smoking, obesity and diabetes, additional research on the combined effect of these health risk factors is warranted.

Keywords: prostatic neoplasms, smoking, prostate-specific antigen, early detection of cancer


Although the mortality benefit from screening with PSA is uncertain,1,2 PSA based screening has resulted in a downward stage shift to more organ confined prostate cancer at the time of diagnosis.3 The often used PSA cut point of greater than 4.0 ng/ml detects the majority of prostate cancers. However, the specificity of the PSA test is low and has led to a high rate of false-positive results, particularly among older men with BPH. Percent-free PSA (the ratio of fPSA-to-tPSA) testing has been used as an adjunct to PSA tests to enhance the specificity and decrease the number of unnecessary biopsies.4

It has been recognized that age, race, BPH, prostate size and prostatitis can influence PSA levels.5,6 Epidemiological studies have also shown that PSA can be affected by several heath characteristics including obesity, diabetes and certain medications such as NSAIDs, statins and thiazide diuretics.79 Scant data exist on the association between smoking and PSA levels. Gelmann et al reported that current and former smokers had a statistically significantly lower PSA and fPSA levels compared with never smokers.10 Gray et al measured PSA values in New Zealand Europeans, Maori and Pacific Islanders, and concluded that this inverse association was only true for fPSA but not tPSA.11 Kristal et al reported that current smokers had lower levels of tPSA than nonsmokers and this inverse association was borderline significant.12 While provocative, these studies produced inconsistent results because potential confounders such as diabetes, BPH and the use of certain medications were not controlled for.

Smoking has been a devastating health concern for decades. In 2010 the U.S. Centers for Disease Control and Prevention reported that nearly a quarter of American men 18 years old or older, approximately 23 million, were current cigarette smokers in 2009.13 Given the high prevalence of smoking and the frequency of PSA screening, it is important to examine the associations between smoking and PSA level while controlling for key sociodemographic and clinical variables using nationally representative data. Our study may aid in interpreting PSA values more accurately, and may have possible implications for the use of tPSA and %fPSA in recommending further evaluation of abnormal PSA results with a prostate biopsy.

MATERIALS AND METHODS

Study Population

The NHANES is a nationally representative cross-sectional survey of the adult United States noninstitutionalized population. A detailed description of sampling and data collection procedures has been published elsewhere.14 The overall response rates for the 2001 to 2002, 2003 to 2004 and 2005 to 2006 surveys were 84%, 79% and 80% for the interview component, and 80%, 76% and 77% for the examination component, respectively.

Men 40 years old or older were eligible for PSA testing consisting of measured tPSA and fPSA as well as a calculated %fPSA. For the 2001–2006 NHANES 4,869 men were 40 years old or older, of whom 4,589 (94.2%) participated in the NHANES physical examination. Men were excluded from PSA testing if they reported current infection or inflammation of the prostate gland, digital rectal examination in the last week, prostate biopsy in the last 30 days, cystoscopy in the last 30 days, history of prostate cancer or if information was missing for any of those questions (323, 7.0%). In addition, 339 (7.4%) men refused or did not give permission for the PSA test and 107 (2.3%) eligible men were missing PSA values. The final study population included 3,820 of 4,589 or 83.2% of all examined men 40 years old or older who participated in the NHANES PSA testing.

Statistical Analysis

Smoking status was defined using cotinine level, and the 2 survey questions, “Have you smoked at least 100 cigarettes in your entire life?” and “Do you now smoke cigarettes every day, some days or not at all?” Serum cotinine levels greater than 10 ng/ml are associated with active smoking within the last few days.15 Respondents with a serum cotinine concentration greater than 10 ng/ml or who had smoked at least 100 cigarettes during their lifetime and, at the time of interview, reported smoking every day or some days were classified as current smokers. Those who responded “not at all” to the latter question were classified as former smokers. Finally, respondents who had smoked fewer than 100 cigarettes in their lifetime were classified as never smokers. Distributions of tPSA, fPSA and %fPSA for the 2001–2006 NHANES were estimated by smoking status as well as other demographic and clinical characteristics. Of those characteristics race/ethnicity was categorized as nonHispanic white, non-Hispanic black, Mexican-American (Mexican or Mexican-American) and other (other Hispanic or other race including multiracial), and tPSA and fPSA were analyzed as continuous variables. In this study medications included statins, thiazide diuretics and NSAIDs. Over the counter NSAIDs were not examined because the NHANES questionnaires did not collect this information. In the descriptive analysis %fPSA was dichotomized to less than 25% vs 25% or greater since less than 25% is typically the cut point for further diagnostic evaluation. Continuous variables were presented as weighted medians and discrete variables as weighted percentages. The 95% CIs for the medians were determined using the Woodruff method.16 For discrete variables 95% CIs were calculated using a logit transformation. No statistical tests were conducted in the bivariate analysis because statistical testing for differences in medians was not available in software packages for complex sample survey data. CIs around the median were included to provide estimates of the variability and allowed for informal comparisons.

Multivariate linear regression models were fit to determine the adjusted relationship between smoking status and PSA values (tPSA and %fPSA). Independent variables included age, race, BMI, diabetes, BPH and C-reactive protein. We included those medications mentioned in the model because previously published studies showed possible associations between these medications and PSA.7,17 We excluded a family history of prostate cancer because this information was only included in the 2003–2006 NHANES. The linearity assumption for continuous predictors was assessed in each model using restricted cubic spline functions.18 The relationship between C-reactive protein and each PSA outcome was nonlinear, and was transformed using a 3-knot restricted cubic spline function. The relationships between the other continuous predictors and each PSA outcome were found to be linear. As such, age and BMI were treated as linear effects in the final models. The dependent variable, tPSA, was log transformed because of nonnormality. Due to the small number of the denominator df, the F-statistic with Satterthwaite correction for the df was used to test significance in each model. Since tPSA was log transformed, the adjusted associations were presented as the percent change in the tPSA geometric mean. The adjusted associations with %fPSA were presented as the absolute change in %fPSA.

Variance estimates were calculated using Taylor series linearization. Sample weights, which account for the differential probabilities of selection, nonresponse and noncoverage, were incorporated into the variance estimation process. The R Package for Statistical Computing was used to estimate the percentiles and corresponding CIs using the Woodruff method.19 SAS-Callable SUDAAN® version 9.0 and SAS® version 9.2 were used for the remainder of the analyses including the regression modeling.

RESULTS

For all ages combined the median (95% CI) tPSA and fPSA was 0.90 (0.81– 0.90) ng/ml and 0.26 (0.25–0.28) ng/ml, respectively (table 1). tPSA and fPSA increased with age and decreased with increasing BMI. Lower values of tPSA were observed for current smokers (0.80 [0.80–0.90] ng/ml) than for former smokers (0.95 [0.83–1.01] ng/ml). The fPSA in current smokers (0.24 [0.23– 0.25] ng/ml) was lower than that in former smokers (0.28 [0.27–0.30] ng/ml) and never smokers (0.27 [0.25– 0.29] ng/ml). Men with BPH had higher tPSA and fPSA levels than those without BPH. Overall 34.2% of men had a %fPSA less than 25%. Current smokers had a higher %fPSA less than 25% (38.3%) than former smokers (31.3%) and never smokers (33.1%). Diabetic men had a higher %fPSA less than 25% than nondiabetic men. Interestingly we found that %fPSA less than 25% increased with increasing C-reactive protein.

Table 1.

PSA levels by study population characteristics

Sample Size Median ng/ml tPSA (95% CI) Median ng/ml fPSA (95% CI) Median %fPSA Less Than 25% (95% CI)
Total 3,820 0.90 (0.81–0.90) 0.26 (0.25–0.28) 34.2 (31.6–36.9)
Age:
  40–49 1,115 0.70 (0.70–0.75) 0.21 (0.20–0.23) 31.8 (28.5–35.3)
  50–59 827 0.90 (0.80–0.95) 0.26 (0.24–0.27) 35.3 (31.9–38.9)
  60–69 840 1.20 (1.08–1.35) 0.33 (0.31–0.36) 39.2 (34.8–43.8)
  70+ 1,038 1.70 (1.58–1.90) 0.48 (0.45–0.52) 32.8 (28.2–37.9)
Race/ethnicity:
  NonHispanic white 2,205 0.90 (0.80–0.94) 0.27 (0.25–0.28) 33.6 (30.9–36.5)
  NonHispanic black 709 0.90 (0.80–1.00) 0.27 (0.25–0.30) 34.0 (29.9–38.4)
  Mexican-American 695 0.90 (0.80–0.97) 0.24 (0.22–0.25) 40.1 (34.8–45.7)
  Other 211 0.80 (0.70–1.00) 0.24 (0.20–0.28) 36.7 (26.3–48.4)
Family history of prostate Ca:*
  No 2,073 0.85 (0.80–0.90) 0.26 (0.25–0.27) 31.7 (28.8–34.7)
  Yes 199 0.97 (0.73–1.30) 0.31 (0.25–0.35) 31.0 (23.9–39.1)
Smoking status:
  Current smoker 1,188 0.80 (0.80–0.90) 0.24 (0.23–0.25) 38.3 (34.6–42.1)
  Former smoker 1,359 0.95 (0.83–1.01) 0.28 (0.27–0.30) 31.3 (27.3–35.6)
  Never smoker 1,270 0.90 (0.80–0.94) 0.27 (0.25–0.29) 33.1 (29.4–37.0)
BMI (kg/m2):
  Normal (less than 25) 905 1.00 (0.90–1.10) 0.30 (0.28–0.33) 35.2 (31.2–39.3)
  Overweight (25–less than 30) 1,621 0.90 (0.80–0.95) 0.27 (0.25–0.28) 33.5 (30.2–36.9)
  Obese (30+) 1,187 0.80 (0.76–0.89) 0.25 (0.23–0.25) 34.3 (30.2–38.6)
Diabetes:
  No 3,253 0.90 (0.84–0.91) 0.26 (0.25–0.27) 35.0 (32.4–37.7)
  Yes 564 0.80 (0.73–0.90) 0.27 (0.25–0.29) 27.7 (21.7–34.5)
BPH:
  No 3,114 0.86 (0.80–0.90) 0.26 (0.25–0.27) 33.7 (30.9–36.6)
  Yes 468 1.27 (1.00–1.50) 0.35 (0.33–0.38) 37.9 (31.5–44.9)
Statin use:
  No 2,900 0.90 (0.80–0.90) 0.26 (0.25–0.27) 34.4 (31.3–37.7)
  Yes 769 0.97 (0.83–1.00) 0.28 (0.26–0.30) 33.6 (29.4–38.1)
NSAID use:
  No 3,419 0.90 (0.82–0.90) 0.26 (0.25–0.27) 34.2 (31.6–36.9)
  Yes 401 0.90 (0.74–1.00) 0.28 (0.26–0.30) 34.4 (27.8–41.7)
Thiazide diuretic use:
  No 3,426 0.90 (0.80–0.90) 0.26 (0.25–0.27) 34.9 (32.2–37.6)
  Yes 394 1.00 (0.82–1.10) 0.29 (0.25–0.33) 27.7 (21.1–35.5)
Cotinine level (ng/ml):
  10 or Less 2,668 0.90 (0.86–1.00) 0.28 (0.26–0.29) 32.7 (29.8–35.8)
  Greater than 10 1,142 0.80 (0.78–0.90) 0.24 (0.22–0.25) 37.6 (33.8–41.6)
C-reactive protein (mg/dl):
  Less than 1.0 3,499 0.90 (0.80–0.90) 0.26 (0.25–0.27) 33.6 (31.0–36.3)
  1.0–3.0 261 0.84 (0.73–1.15) 0.25 (0.22–0.31) 40.1 (31.9–48.9)
  Greater than 3.0 60 1.16 (1.00–1.40) 0.30 (0.21–0.40) 53.7 (36.3–70.3)
*

Only available for the NHANES 2003–2006.

Prescription only.

The results of the multivariate linear regression model predicting tPSA are shown in table 2. The predicted geometric mean of tPSA was 7.9% and 12.2% lower among current and former smokers, respectively, compared with never smokers. A 5-year increase in age was associated with a significant 13.7% increase in the geometrical mean of tPSA. However, a 5 kg/m2 increase in BMI was associated with a significant 6.9% decrease in tPSA. Diabetes was associated with a 14.8% decrease in tPSA compared with nondiabetics, while BPH was associated with a 16.5% increase in tPSA. The predicted geometric mean of tPSA was 14.1% higher among non-Hispanic black vs nonHispanic white men. Comparing medication use and the geometric mean of tPSA, no statistically significant association was found. C-reactive protein had a significant nonlinear relationship with tPSA.

Table 2.

Multivariate model predicting geometric mean of tPSA

% Change in Predicted Geometric
Mean (95% CI)
p Value*
Age (per 5-yr increase) 13.7 (12.1, 15.4) <0.001
Race/ethnicity: 0.098
  NonHispanic white 0 (reference)
  NonHispanic black 14.1 (3.8, 25.3)
  Mexican-American 10.1 (−0.7, 22.1)
  Other 0.4 (−13.7, 16.7)
Smoking status: 0.015
  Current smoker −7.9 (−15.1, −0.1)
  Former smoker −12.2 (−20.5, −3.0)
  Never smoker 0 (reference)
BMI (per 5 kg/m2 increase) −6.9 (−9.3, −4.4) <0.001
Diabetes: 0.003
  No 0 (reference)
  Yes −14.8 (−23.2, −5.4)
BPH: 0.016
  No 0 (reference)
  Yes 16.5 (3.1, 31.7)
Statin use: 0.274
  No 0 (reference)
  Yes −4.3 (−11.6, 3.6)
NSAID use: 0.217
  No 0 (reference)
  Yes −6.3 (−15.6, 4.0)
Thiazide diuretic use: 0.470
  No 0 (reference)
  Yes −4.6 (−16.1, 8.6)
C-reactive protein Nonlinear (nonlinear) 0.001

All variables included in the model are shown in the table.

*

Based on the F-statistic with Satterthwaite correction for the degrees of freedom from the simultaneous test that all coefficients for a given variable are equal to 0.

Prescription only.

Increasing C-reactive protein was associated with higher tPSA for C-reactive protein values up to 0.6 mg/dl. Above this level C-reactive protein did not have a significant effect on tPSA.

Results of the multivariate linear regression predicting %fPSA are presented in table 3. Current smokers had a statistically significantly lower %fPSA compared with former smokers (change in %fPSA −2.4; 95% CI −3.8, −0.9; comparison not shown). Diabetes was associated with a higher %fPSA (change in %fPSA 3.8; 95% CI 1.4, 6.2) while increased age was associated with a significant decrease in %fPSA (change in %fPSA −0.3; 95% CI −0.5, −0.05 for 5-year increase). Mexican-Americans had a lower %fPSA compared with nonHispanic whites (change in %fPSA −3.2; 95% CI −4.8, −1.7). There was also a significant nonlinear relationship between C-reactive protein and %fPSA.

Table 3.

Multivariate model predicting % fPSA

Change in Predicted %fPSA
Mean (95% CI)
p Value*
Age (per 5-yr increase) −0.3 (−0.5, −0.05) 0.021
Race/ethnicity: 0.048
  NonHispanic white 0 (reference)
  NonHispanic black −0.4 (−1.8, 0.9)
  Mexican-American −3.2 (−4.8, −1.7)
  Other −1.3 (−4.4, 1.7)
Smoking status: 0.015
  Current smoker −1.2 (−2.8, 0.4)
  Former smoker 1.2 (−0.1, 2.5)
  Never smoker 0 (reference)
BMI (per 5 kg/m2 increase) 0.0 (−0.6, 0.5) 0.926
Diabetes: 0.003
  No 0 (reference)
  Yes 3.8 (1.4, 6.2)
BPH: 0.145
  No 0 (reference)
  Yes −1.4 (−3.4, 0.5)
Statin use: 0.812
  No 0 (reference)
  Yes −0.2 (−1.6, 1.2)
NSAID use: 0.287
  No 0 (reference)
  Yes 1.2 (−1.0, 3.4)
Thiazide diuretic use: 0.151
  No 0 (reference)
  Yes 1.5 (−0.6, 3.7)
C-reactive protein Nonlinear (nonlinear) <0.001

All variables included in the model are shown in the table.

*

Based on the F-statistic with Satterthwaite correction for the degrees of freedom from the simultaneous test that all coefficients for a given variable are equal to 0.

Prescription only.

Increasing C-reactive protein was associated with a decrease in %fPSA for levels up to 0.7 mg/dl. Above this level there was no relationship between C-reactive protein and %fPSA.

DISCUSSION

Using nationally representative data and adjusting for multiple confounders, we found that current and former smokers had a statistically significantly lower tPSA compared with never smokers, and current smokers had a significantly lower %fPSA than former smokers. Gelmann et al examined the relationship between smoking and PSA using the study cohort of the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial.10 The authors found that tPSA and fPSA levels were significantly lower among current and former smokers, respectively, compared with never smokers. Similarly an analysis of the 2001–2002 NHANES data demonstrated that men who had ever smoked had a lower PSA than those who had not.17 The underlying causes for the inverse associations between smoking and tPSA levels are largely unknown. It has been reported that male smokers had an increased level of circulating sex hormone-binding globulin, which might reduce the synthesis of PSA.20 To our knowledge our study finding that current smokers had a significantly lower %fPSA than former smokers has not been reported in the literature. Additional research is needed to verify this finding.

To date the U.S. Preventive Services Task Force does not recommend or advise against routine screening for prostate cancer for men 50 to 74 years old.21,22 The American Urological Association recommends starting a discussion about prostate cancer screening with men who are at average risk at age 40 years to establish a baseline PSA.23 While adjusting for other potential confounders, our study showed that there was an approximate 8% to 12% decrease in tPSA among current and former smokers. Thus, men who have ever smoked are less likely to have an abnormal result on PSA screening and diagnostic biopsy, possibly resulting in fewer screen detected prostate cancers than in those who have never smoked. In this study we excluded men with prostate cancer, prostatic infection or inflammation. In this relatively healthy population the median tPSA was 0.9 ng/ml. Loeb et al showed that a baseline PSA between 0.7 and 2.5 ng/ml in men 40 to 49 years old was associated with a 14.6-fold increased risk of prostate cancer.24 In addition to the often used PSA cut point of greater than 4.0 ng/ml, age specific reference ranges for PSA have been demonstrated as a useful approach to improve PSA sensitivity and specificity.25

A survival analysis using data from the National Institutes of Health-American Association of Retired Persons Diet and Health Study suggested that current and former smokers might be at decreased risk for prostate cancer diagnosis, primarily localized disease.26 A meta-analysis of 24 cohort studies also suggested that current smokers were not at increased risk for incident prostate cancer.27 These findings may be partially attributable to detection bias resulting from the inverse relationship between smoking and PSA. If smoking substantially alters the relationship between PSA and any potential underlying prostate cancer, we may need to modify, according to smoking status, the often used PSA cutoff points that signal the necessity of further evaluation with prostate biopsy. Such a modification may help to interpret screening results more accurately.

Biologically the mechanisms of initiating or accelerating prostate cancer by smoking have been postulated, including exposure to carcinogenic compounds, modulation of male hormones and genetic mutations in suppressor genes.28 Despite the demonstrated links between smoking and several types of solid tumors, the role of smoking in the development of prostate cancer is inconclusive. However, studies have consistently shown that current smokers have an increased risk of fatal prostate cancer.27,29 Delayed detection of prostate cancer due to the effects of smoking on PSA may contribute in part to worse health outcomes of prostate cancer among smokers.

It has been widely reported that obesity and diabetes have inverse associations with PSA levels.8,9,11 These findings were also confirmed by our study. Interestingly we did not find any statistically significant association between PSA and the use of NSAIDs, statins or thiazide diuretics as found by Chang et al.7 Different ways of defining smoking status, and various sociodemographic, clinical factors and medications examined in these 2 studies may partially account for this difference. Although inverse associations between PSA and the use of NSAIDs and statins have been reported,7,17 a recent study suggested that use of NSAIDs (excluding aspirin) and statins was not associated with PSA, and only aspirin users had a statistically significantly lower PSA than nonusers among men who had never smoked.30

A major strength of this study is that it was based on nationwide, population based sampling survey data. However, it has at least 3 limitations. 1) Smoking status was defined based on self-report and serum cotinine levels. Self-report smoking status was subject to misclassification errors. The use of cotinine level in defining current smokers should help reduce the misclassification of self-reported smoking status. 2) This study was cross-sectional, and the temporal relationship of associations between smoking and PSA cannot be assessed. 3) The NHANES did not collect information on nonprescription medication use so the effects of any nonprescription medications cannot be examined.

CONCLUSIONS

Men who had ever smoked had an 8% to 12% decrease in tPSA compared with those who had never smoked. Although the magnitude of the inverse association between smoking status and PSA alone may not be substantial, the joint effect of smoking, obesity and diabetes associated with PSA may have greater clinical implications given the high prevalence of these conditions. Additional research on the combined effect of these and other health risk factors on PSA is needed. Further studies to better understand the biological role of smoking in the development of prostate cancer are also warranted.

Acknowledgments

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Abbreviations and Acronyms

%fPSA

percent free prostate specific antigen

BMI

body mass index

BPH

benign prostatic hyperplasia

fPSA

free prostate specific antigen

NSAIDs

nonsteroidal anti-inflammatory drugs

PSA

prostate specific antigen

tPSA

total prostate specific antigen

REFERENCES

  • 1.Andriole GL, Crawford ED, Grubb RL, 3rd, et al. Mortality results from a randomized prostate-cancer screening trial. N Engl J Med. 2009;360:1310. doi: 10.1056/NEJMoa0810696. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Schröder FH, Hugosson J, Roobol MJ, et al. Screening and prostate-cancer mortality in a randomized European study. N Engl J Med. 2009;360:1320. doi: 10.1056/NEJMoa0810084. [DOI] [PubMed] [Google Scholar]
  • 3.Stephenson RA. Prostate cancer trends in the era of prostate-specific antigen. An update of incidence, mortality, and clinical factors from the SEER database. Urol Clin North Am. 2002;29:173. doi: 10.1016/s0094-0143(02)00002-2. [DOI] [PubMed] [Google Scholar]
  • 4.Partin AW, Catalona WJ, Southwick PC, et al. Analysis of percent free prostate-specific antigen (PSA) for prostate cancer detection: influence of total PSA, prostate volume, and age. Urology. 1996;48:55. doi: 10.1016/s0090-4295(96)00611-5. [DOI] [PubMed] [Google Scholar]
  • 5.Oesterling JE, Jacobsen SJ, Chute CG, et al. Serum prostate-specific antigen in a community-based population of healthy men. Establishment of age-specific reference ranges. JAMA. 1993;270:860. [PubMed] [Google Scholar]
  • 6.Eastham JA, May RA, Whatley T, et al. Clinical characteristics and biopsy specimen features in African-American and white men without prostate cancer. J Natl Cancer Inst. 1998;90:756. doi: 10.1093/jnci/90.10.756. [DOI] [PubMed] [Google Scholar]
  • 7.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. 2010;28:3951. doi: 10.1200/JCO.2009.27.9406. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Werny DM, Saraiya M, Gregg EW. Prostate-specific antigen values in diabetic and nondiabetic US men, 2001–2002. Am J Epidemiol. 2006;164:978. doi: 10.1093/aje/kwj311. [DOI] [PubMed] [Google Scholar]
  • 9.Werny DM, Thompson T, Saraiya M, et al. Obesity is negatively associated with prostate-specific antigen in U.S. men, 2001–2004. Cancer Epidemiol Biomarkers Prev. 2007;16:70. doi: 10.1158/1055-9965.EPI-06-0588. [DOI] [PubMed] [Google Scholar]
  • 10.Gelmann EP, Chia D, Pinsky PF, et al. Relationship of demographic and clinical factors to free and total prostate-specific antigen. Urology. 2001;58:561. doi: 10.1016/s0090-4295(01)01305-x. [DOI] [PubMed] [Google Scholar]
  • 11.Gray MA, Delahunt B, Fowles JR, et al. Demographic and clinical factors as determinants of serum levels of prostate specific antigen and its derivatives. Anticancer Res. 2004;24:2069. [PubMed] [Google Scholar]
  • 12.Kristal AR, Chi C, Tangen CM, et al. Associations of demographic and lifestyle characteristics with prostate-specific antigen (PSA) concentration and rate of PSA increase. Cancer. 2006;106:320. doi: 10.1002/cncr.21603. [DOI] [PubMed] [Google Scholar]
  • 13.Centers for Disease Control and Prevention (CDC) Vital signs: current cigarette smoking among adults aged >or=18 years — United States, 2009. MMWR Morb Mortal Wkly Rep. 2010;59:1135. [PubMed] [Google Scholar]
  • 14.Joseph DA, Thompson T, Saraiya M, et al. Association between glomerular filtration rate, free, total, and percent free prostate-specific antigen. Urology. 2010;76:1042. doi: 10.1016/j.urology.2009.05.100. [DOI] [PubMed] [Google Scholar]
  • 15.Centers for Disease Control and Prevention. Fourth National Report on Human Exposure to Environmental Chemicals 2009. [Accessed May 6, 2011]; Available at http://www.cdc.gov/exposurereport.
  • 16.Woodruff RS. Confidence intervals for medians and other positional measures. J Am Stat Assoc. 1952;47:635. [Google Scholar]
  • 17.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;113:2053. doi: 10.1002/cncr.23806. [DOI] [PubMed] [Google Scholar]
  • 18.Harrell FE., Jr . Regression Modeling Strategies, With Applications to Linear Models, Logistic Regression, and Survival Analysis. New York: Springer; 2001. [Google Scholar]
  • 19.R Development Core Team: R. A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing; 2006. [Google Scholar]
  • 20.Field AE, Colditz GA, Willett WC, et al. The relation of smoking age, relative weight, and dietary intake to serum adrenal steroids, sex hormones, and sex hormone-binding globulin in middle-aged men. J Clin Endocrinol Metab. 1994;79:1310. doi: 10.1210/jcem.79.5.7962322. [DOI] [PubMed] [Google Scholar]
  • 21.U.S. Preventive Services Task Force. Screening for prostate cancer: recommendation and rationale. Ann Intern Med. 2002;137:915. doi: 10.7326/0003-4819-137-11-200212030-00013. [DOI] [PubMed] [Google Scholar]
  • 22.U.S. Preventive Services Task Force. Screening for prostate cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2008;149:185. doi: 10.7326/0003-4819-149-3-200808050-00008. [DOI] [PubMed] [Google Scholar]
  • 23.Greene KL, Albertsen PC, Babaian RJ, et al. Prostate specific antigen best practice statement: 2009 update. J Urol. 2009;182:2232. doi: 10.1016/j.juro.2009.07.093. [DOI] [PubMed] [Google Scholar]
  • 24.Loeb S, Roehl KA, Antenor JA, et al. Baseline prostate-specific antigen compared with median prostate-specific antigen for age group as predictor of prostate cancer risk in men younger than 60 years old. Urology. 2006;67:316. doi: 10.1016/j.urology.2005.08.040. [DOI] [PubMed] [Google Scholar]
  • 25.Loeb S, Nadler RB, Roehl KA, et al. Risk of prostate cancer for young men with a prostate specific antigen less than their age specific median. J Urol. 2007;177:1745. doi: 10.1016/j.juro.2007.01.068. [DOI] [PubMed] [Google Scholar]
  • 26.Watters JL, Park Y, Hollenbeck A, et al. Cigarette smoking and prostate cancer in a prospective US cohort study. Cancer Epidemiol Biomarkers Prev. 2009;18:2427. doi: 10.1158/1055-9965.EPI-09-0252. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Huncharek M, Haddock KS, Reid R, et al. Smoking as a risk factor for prostate cancer: a meta-analysis of 24 prospective cohort studies. Am J Public Health. 2010;100:693. doi: 10.2105/AJPH.2008.150508. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Hickey K, Do KA, Green A. Smoking and prostate cancer. Epidemiol Rev. 2001;23:115. doi: 10.1093/oxfordjournals.epirev.a000776. [DOI] [PubMed] [Google Scholar]
  • 29.Kenfield SA, Stampfer MJ, Chan JM, et al. Smoking and prostate cancer survival and recurrence. JAMA. 2011;305:2548. doi: 10.1001/jama.2011.879. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Algotar AM, Thompson PA, Ranger-Moore J, et al. Effect of aspirin, other NSAIDs, and statins on PSA and PSA velocity. Prostate. 2010;70:883. doi: 10.1002/pros.21122. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES