bs_bs_banner
International Journal of Urology (2014) 21, 987–990
doi: 10.1111/iju.12486
Original Article: Clinical Investigation
Impact of obesity on the predictive accuracy of prostate-specific antigen density and prostate-specific antigen in native Korean men undergoing prostate biopsy Jae Heon Kim,1 Seung Whan Doo,1 Won Jae Yang,1 Kwang Woo Lee,1 Chang Ho Lee,1 Yun Seob Song,1 Yoon Su Jeon,1 Min Eui Kim1 and Soon-Sun Kwon2 1
Department of Urology, College of Medicine, Soonchunhyang University, Cheonan, and 2Medical Research Collaborating Center, Seoul National University Bundang Hospital, Seongnam, South Korea
Abbreviations & Acronyms AUC = area under the curve BMI = body mass index DRE = digital rectal examination PCa = prostate cancer PSA = prostate-specific antigen PSAD = prostate-specific antigen density PV = prostate volume ROC = receiver operating characteristic TRUS = transrectal ultrasonography Correspondence: Won Jae Yang M.D., Ph.D., Department of Urology, Soonchunhyang University, Seoul Hospital, 59 Daesagwan-ro, Yongsan-gu, Seoul 140-743, Korea. Email: [email protected] Received 7 January 2014; accepted 6 April 2014. Online publication 13 May 2014
© 2014 The Japanese Urological Association
Objectives: To evaluate the impact of obesity on the biopsy detection of prostate cancer. Methods: We retrospectively reviewed data of 1182 consecutive Korean patients (≥50 years) with serum prostate-specific antigen levels of 3–10 ng/mL who underwent initial extended 12-cores biopsy from September 2009 to March 2013. Patients who took medications that were likely to influence the prostate-specific antigen level were excluded. Receiver operating characteristic curves were plotted for prostate-specific antigen and prostatespecific antigen density predicting cancer status among non-obese and obese men. Results: A total of 1062 patients (mean age 67.1 years) were enrolled in the analysis. A total of 230 men (21.7%) had a positive biopsy. In the overall study sample, the area under the receiver operator characteristic curve of serum prostate-specific antigen for predicting prostate cancer on biopsy were 0.584 and 0.633 for non-obese and obese men, respectively (P = 0.234). However, the area under the curve for prostate-specific antigen density in predicting cancer status showed a significant difference (non-obese 0.696, obese 0.784; P = 0.017). Conclusions: There seems to be a significant difference in the ability of prostate-specific antigen density to predict biopsy results between non-obese and obese men. Obesity positively influenced the overall ability of prostate-specific antigen density to predict prostate cancer.
Key words: obesity, prostatic specific antigen, prostatic specific antigen density, receiver operating characteristic curve.
Introduction Obese men are known to have a significantly higher rate of PCa deaths, and have lower serum PSA concentrations compared with non-obese men.1,2 It has been suggested that obese men experience a PCa detection bias as a result of hemodilution of PSA that can result in delayed PCa diagnosis and subsequent delay in treatment.3 If obesity leads to a clinically significant decrease in PSA, this would suggest that the PSA thresholds used in clinical practice should be adjusted based on BMI. However, obesity also has a significant positive correlation with PV, and the larger the PV, the higher the PSA.4–6 Therefore, obesity has a conflicting influence on PSA level. PSAD was initially devised to optimize the clinical use of PSA in an attempt to decrease the number of unnecessary biopsies in men with benign disease.7 We previously reported that PSAD showed a more significant inverse correlation with the degree of obesity than PSA, because it can compensate for the confounding effect of PV on the obesity-related decrease in PSA.8 Although some investigators have found a positive association between obesity and PCa,9–15 others have observed a negative association,16–19 and still others have observed no association whatsoever.20–24 Because of the conflicting aforementioned data, the reliability of PSA as a PCa screening tool in obese men has been questioned. In response to these concerns, we tested the accuracy of PSA and PSAD as a predictor of PCa across increasing BMI categories by comparing PSA and PSAD operating characteristics in a cohort of men who underwent prostate biopsy. 987
JH KIM ET AL.
Methods Study population and data collection After obtaining institutional review board approval, we retrospectively reviewed data on 1182 consecutive Korean patients (age ≥50 years) with serum PSA levels of 3–10 ng/mL who underwent initial extended prostate biopsy from September 2009 to March 2013. Patients who took medications that were likely to influence the PSA level, such as 5α-reductase inhibitor, Serenoa repens or testosterone, before measurement of PSA (112 patients) or with a history of malignant neoplasm in the bladder (8 patients) were excluded. All men underwent detailed clinical evaluations. BMI was defined as weight divided by the square of height (m2). Serum PSA tests were carried out using the automated chemiluminescent microparticle immunoassay analyzer Architect i2000 (Abbott Diagnostic Laboratories, Abbott Park, IL, USA). The prostate was measured in 3-D by TRUS using an 8.0-MHz rectal probe (LOGIQ P6-PRO; GE Healthcare, Little Chalfont, UK), and the PV was estimated using a modification of the prolate ellipsoid formula and recorded in cm3 (0.523 [length (cm) × width (cm) × height (cm)]). PSAD was obtained by calculating the quotient of PSA and PV. Patients underwent TRUS-guided extended biopsies (12 cores). BMI was grouped according to the Asia–Pacific criteria for obesity and were as follows: non-obese (less than 25 kg/m2) versus obese (25 kg/m2 or greater).25
Statistical analyses Measures of central tendency for demographic and clinical parameters across BMI categories were examined for continuous variables, such as age at biopsy, prebiopsy PSA and PV, using independent t-test. Distribution of men in each BMI group across categorical determinants, such as biopsy results (cancer vs non-cancer), and biopsy Gleason group (≤6 vs ≥7) were compared using the χ2-test. A linear regression model controlling for age and PV was used to calculate mean-adjusted PSA, PSAD with 95% CI for each BMI category. ROC curves were plotted as false positive rate (i.e. 1 minus specificity) versus sensitivity for all cut-off values in the range of the PSA values observed. The area under the ROC curve was used to measure the performance accuracy of PSA as a predictor of prostate biopsy results within each BMI category. An AUC of 1.0 represents error-free prediction of cancer status in all samples, whereas an AUC of 0.50 represents a 50% likelihood of a correct prediction of cancer status similar to a coin toss. The AUC were compared across the entire two BMI groups using the DeLong test. All statistics were two-tailed, and a P-value of
International Journal of Urology (2014) 21, 987–990
doi: 10.1111/iju.12486
Original Article: Clinical Investigation
Impact of obesity on the predictive accuracy of prostate-specific antigen density and prostate-specific antigen in native Korean men undergoing prostate biopsy Jae Heon Kim,1 Seung Whan Doo,1 Won Jae Yang,1 Kwang Woo Lee,1 Chang Ho Lee,1 Yun Seob Song,1 Yoon Su Jeon,1 Min Eui Kim1 and Soon-Sun Kwon2 1
Department of Urology, College of Medicine, Soonchunhyang University, Cheonan, and 2Medical Research Collaborating Center, Seoul National University Bundang Hospital, Seongnam, South Korea
Abbreviations & Acronyms AUC = area under the curve BMI = body mass index DRE = digital rectal examination PCa = prostate cancer PSA = prostate-specific antigen PSAD = prostate-specific antigen density PV = prostate volume ROC = receiver operating characteristic TRUS = transrectal ultrasonography Correspondence: Won Jae Yang M.D., Ph.D., Department of Urology, Soonchunhyang University, Seoul Hospital, 59 Daesagwan-ro, Yongsan-gu, Seoul 140-743, Korea. Email: [email protected] Received 7 January 2014; accepted 6 April 2014. Online publication 13 May 2014
© 2014 The Japanese Urological Association
Objectives: To evaluate the impact of obesity on the biopsy detection of prostate cancer. Methods: We retrospectively reviewed data of 1182 consecutive Korean patients (≥50 years) with serum prostate-specific antigen levels of 3–10 ng/mL who underwent initial extended 12-cores biopsy from September 2009 to March 2013. Patients who took medications that were likely to influence the prostate-specific antigen level were excluded. Receiver operating characteristic curves were plotted for prostate-specific antigen and prostatespecific antigen density predicting cancer status among non-obese and obese men. Results: A total of 1062 patients (mean age 67.1 years) were enrolled in the analysis. A total of 230 men (21.7%) had a positive biopsy. In the overall study sample, the area under the receiver operator characteristic curve of serum prostate-specific antigen for predicting prostate cancer on biopsy were 0.584 and 0.633 for non-obese and obese men, respectively (P = 0.234). However, the area under the curve for prostate-specific antigen density in predicting cancer status showed a significant difference (non-obese 0.696, obese 0.784; P = 0.017). Conclusions: There seems to be a significant difference in the ability of prostate-specific antigen density to predict biopsy results between non-obese and obese men. Obesity positively influenced the overall ability of prostate-specific antigen density to predict prostate cancer.
Key words: obesity, prostatic specific antigen, prostatic specific antigen density, receiver operating characteristic curve.
Introduction Obese men are known to have a significantly higher rate of PCa deaths, and have lower serum PSA concentrations compared with non-obese men.1,2 It has been suggested that obese men experience a PCa detection bias as a result of hemodilution of PSA that can result in delayed PCa diagnosis and subsequent delay in treatment.3 If obesity leads to a clinically significant decrease in PSA, this would suggest that the PSA thresholds used in clinical practice should be adjusted based on BMI. However, obesity also has a significant positive correlation with PV, and the larger the PV, the higher the PSA.4–6 Therefore, obesity has a conflicting influence on PSA level. PSAD was initially devised to optimize the clinical use of PSA in an attempt to decrease the number of unnecessary biopsies in men with benign disease.7 We previously reported that PSAD showed a more significant inverse correlation with the degree of obesity than PSA, because it can compensate for the confounding effect of PV on the obesity-related decrease in PSA.8 Although some investigators have found a positive association between obesity and PCa,9–15 others have observed a negative association,16–19 and still others have observed no association whatsoever.20–24 Because of the conflicting aforementioned data, the reliability of PSA as a PCa screening tool in obese men has been questioned. In response to these concerns, we tested the accuracy of PSA and PSAD as a predictor of PCa across increasing BMI categories by comparing PSA and PSAD operating characteristics in a cohort of men who underwent prostate biopsy. 987
JH KIM ET AL.
Methods Study population and data collection After obtaining institutional review board approval, we retrospectively reviewed data on 1182 consecutive Korean patients (age ≥50 years) with serum PSA levels of 3–10 ng/mL who underwent initial extended prostate biopsy from September 2009 to March 2013. Patients who took medications that were likely to influence the PSA level, such as 5α-reductase inhibitor, Serenoa repens or testosterone, before measurement of PSA (112 patients) or with a history of malignant neoplasm in the bladder (8 patients) were excluded. All men underwent detailed clinical evaluations. BMI was defined as weight divided by the square of height (m2). Serum PSA tests were carried out using the automated chemiluminescent microparticle immunoassay analyzer Architect i2000 (Abbott Diagnostic Laboratories, Abbott Park, IL, USA). The prostate was measured in 3-D by TRUS using an 8.0-MHz rectal probe (LOGIQ P6-PRO; GE Healthcare, Little Chalfont, UK), and the PV was estimated using a modification of the prolate ellipsoid formula and recorded in cm3 (0.523 [length (cm) × width (cm) × height (cm)]). PSAD was obtained by calculating the quotient of PSA and PV. Patients underwent TRUS-guided extended biopsies (12 cores). BMI was grouped according to the Asia–Pacific criteria for obesity and were as follows: non-obese (less than 25 kg/m2) versus obese (25 kg/m2 or greater).25
Statistical analyses Measures of central tendency for demographic and clinical parameters across BMI categories were examined for continuous variables, such as age at biopsy, prebiopsy PSA and PV, using independent t-test. Distribution of men in each BMI group across categorical determinants, such as biopsy results (cancer vs non-cancer), and biopsy Gleason group (≤6 vs ≥7) were compared using the χ2-test. A linear regression model controlling for age and PV was used to calculate mean-adjusted PSA, PSAD with 95% CI for each BMI category. ROC curves were plotted as false positive rate (i.e. 1 minus specificity) versus sensitivity for all cut-off values in the range of the PSA values observed. The area under the ROC curve was used to measure the performance accuracy of PSA as a predictor of prostate biopsy results within each BMI category. An AUC of 1.0 represents error-free prediction of cancer status in all samples, whereas an AUC of 0.50 represents a 50% likelihood of a correct prediction of cancer status similar to a coin toss. The AUC were compared across the entire two BMI groups using the DeLong test. All statistics were two-tailed, and a P-value of
