| OBJECTIVE (1) To develop a nomogram for predicting the probability of a positive initial prostate biopsy in a Chinese population.(2) To assess the efficiencies of transition zone prostate-specific antigen (PSA) density (TZPSAD) in the diagnosis of prostate cancer in Chinese men with PSA of both4.0-10.0and10.1-20.0ng/ml.(3) To explore the distributions and characteristics of initial PSA and PSA velocity in men≤50years old without prostate cancer.(4) To evaluate the association between prostate volume (PV) and prostate cancer detection rate in men with a PSA of10-50ng/ml.(5) To evaluate the efficiency of free/total (f/t) PSA in diagnosis of prostate cancer in Chinese men with PSA4.0-10.0and10.1-20.0ng/ml.MATERIALS AND METHODS (1) A total of699men who had undergone trans-rectal ultrasound (TRUS)-guided prostate biopsy for the detection of cancer between November1999and June2011were retrieved from our hospital. Twelve men who were biopsied more than2times and152men without complete data were excluded. This left a total of535subjects who were included in the study. Within this cohort,486(90.8%) underwent a13-core biopsy,28(5.2%) underwent a6-core biopsy, and21(3.9%) underwent a4-15core biopsy (not a6-or13-core).PSA was analysed with the Abbott AxSYM PSA (Abbott Corporation) assay before2006and the Roche Elecsys2010PSA assay (Roche Diagnostics) after2006. Prostate volume (PV) was determined during TRUS and was calculated using the formula, PV (ml)=0.52×anterior-posterior diameter (cm)×transverse diameter (cm)×superior-inferior diameter (cm). Age, PSA, PV, digital rectal examination (DRE) status, f/t PSA and TRUS findings (low-echogenicity in the peripheral zone was defined as positive) were included in the stepwise regression analysis. Variables that did not reach statistical significance at a level of0.05were removed from the model in the backward stepwise process. The independent variables from the final model were used to construct a nomogram to predict the probability of a positive initial biopsy. The nomogram was validated and predictive accuracy calculated using Harrell’s Concordance Index C on100bootstrapped re-samples. A Receiver operating characteristic (ROC) curve was used to evaluate the effectiveness of the nomogram and PSA alone in predicting a positive initial prostate biopsy. Statistical analyses were performed with R version2.13.1(http://www.r-project.org/).(2) A total of607men who had undergone trans-rectal ultrasound-guided systematic prostatic biopsy for detecting prostate cancer from November1999to August2009were retrieved from our hospital. Of the607men who had a prostatic biopsy,384had a document of trans-rectal ultrasound transition zone and PSA measurements. A total of189men with a biopsy PSA of4.0-20.0ng/ml were included in the study. Of these men,183,5,1had undergone13,6, and14systematic cores prostatic biopsy, respectively. PSA measurement and prostatic biopsy were performed as Chapter (1). Prostate transition zone volume (TZV) was calculated using the following formula:TZV (ml)=0.52×anterior-posterior diameter (cm)×transverse diameter (cm)×superior-inferior diameter (cm). TZPSAD (ng/ml/ml)=PSA (ng/ml)÷TZV (ml). A ROC curve was used to compare the efficiency of PSA and TZPSAD in the diagnosis of prostate cancer in men with a PSA of4.0-10.0and10.1-20.0ng/ml.The efficiencies in the diagnosis of prostate cancer for different PSA and TZPSAD cut-offs in men with PSA4.0-10.0and10.1-20.0ng/ml were also evaluated. Total efficiency in the diagnosis of prostate cancer for different PSA and TZPSAD cut-offs was calculated using the following formula:total efficiency=sensitivity x specificity. The cut-off of the greatest efficiency on the ROC curve was considered as best cut-off for diagnosis of prostate cancer. Student-t test was used to compare the differences of age, PSA, TZV, and TZPSAD between prostate cancer and non-prostate cancer group. Chi square test was used to compare the detection rate of prostate cancer between PSA4.0-10.0and10.1-20.0ng/ml group. Statistical analyses were performed using SPSS program (version17.0, Chicago, IL). All statistical tests were two-sided with a P<0.05considered statistically significant.(3) A total of4211men had their initial PSA measurement at age≤50years old from January2001to November2009were retrospectively retrieved from our hospital. Of these men,11had undergone prostatic biopsy and5were diagnosed with prostate cancer. Those with a diagnosis of prostate cancer were excluded. This left4206men were included in the study. Of these men, PSA velocity was calculated in417who had their PSA measurement twice or more. PSA measurement and prostatic biopsy were performed as Chapter (1). PSA velocity was calculated as the slope of linear regression line of all PSA values collected over time. The important cutoffs of baseline PSA (1.0,2.5and4.0ng/ml) and PSA velocity (0.35,0.75,1.0ng/ml/year) were used to stratify patients in different age groups (≤30,31-39, and40-50years old). The differences of baseline PSA and PSA velocity and the correlation of baseline PSA and PSA velocity were assessed. The distributions of baseline PSA and PSA velocity in men age≤50years old were demonstrated. The correlations between initial PSA, initial PSA age, and PSA velocity were also estimated. Kaplan-meier and log-rank test were used to estimate the significant difference at the risk of PSA≥2.5ng/ml after initial PSA measurement, stratified by median initial PSA (0.6ng/ml).(4) A total of699men who had undergone TRUS guided prostate biopsy between November1999and June2011were retrieved from our hospital. Within this cohort,593(84.8%) underwent13-core biopsy,40(5.7%) underwent6-core biopsy and66(9.4%) underwent4-15core biopsy (not6or13cores). Two hundreds and sixty one men with a PSA of10-50ng/ml underwent a13-core prostate biopsy with complete PV data were included in the study. PV was measured via TRUS. PSA measurement, prostatic biopsy, and PV calculation were performed as Chapter (1). The clinical variables included in analysis were age at prostate biopsy, PSA at time of biopsy, PV, and DRE status. PV was used as both a continuous and categorical variable (stratified by median). Age and PSA were analyzed as a continuous variable and DRE was treated as a categorical variable. Multivariate stepwise Logistic regression was used to determine which variables (PV as a continuous or categoric variable, age and PSA as a continuous variable, and DRE as a categoric variable) were predictive of a positive TRUS biopsy. The rates of prostate cancer among men with different DRE statuses and PSA ranges, stratified by PV medians, were calculated. Statistical analyses were performed using SPSS statistical software (version18.0; SPSS Inc,Chicago, IL). All statistical tests were2-sided with a P<0.05considered to be statistically significant.(5) A total of699men who had undergone trans-rectal ultrasound-guided systematic prostatic biopsy for detecting prostate cancer from November1999to June2011were retrieved from our hospital. Of the699men who had a prostatic biopsy,251had a document of free to total PSA measurements with a biopsy PSA of4.0-20.0ng/ml were included in the study. PSA was analyzed with the following assays:PSA measurement and prostatic biopsy were performed as Chapter (1). f/t PSA was calculated using the following formula:f/t PSA=f PSA (ng/ml)÷t PSA (ng/ml). Receiver operating characteristic (ROC) curve was used to compare the efficiency of PSA and f/t PSA in the diagnosis of prostate cancer in men with a PSA of4.0-10.0and10.1-20.0ng/ml. The efficiencies in the diagnosis of prostate cancer for PSA and f/t PSA best cut-offs in men with PSA4.0-10.0and10.1-20.0ng/ml were also evaluated. Total efficiency in the diagnosis of prostate cancer for PSA and f/t PSA best cut-offs was calculated using the following formula:total efficiency=sensitivity x specificity. The cut-off of the greatest efficiency on the ROC curve was considered as best cut-off for diagnosis of prostate cancer. Student-t test was used to compare the differences of age, PSA, PV, and F/T PSA between prostate cancer and non-prostate cancer group. Chi square test was used to compare the detection rate of prostate cancer between PSA4.0-10.0and10.1-20.0ng/ml group. Statistical analyses were performed using SPSS program (version17.0, Chicago, IL). All statistical tests were two-sided with a P<0.05considered statistically significant.RESULT (1) Of the535subjects included in the study,41.7%(223/535) had a positive initial prostate biopsy. The median and mean diagnostic PSA levels in our study cohort were18.6and91.4ng/ml, respectively. The median PSA level in men with a positive initial prostate biopsy was statistically higher than those with a negative biopsy (43.4Vs.13.1ng/ml, P<0.001, Mann-Whitney U test). In a multivariate analysis, only age, PV, LogPSA, and DRE were found to be independent predictors of a positive initial prostate biopsy.The logistic regression model yielded the equation below:Positive initial prostate biopsy probability equation for Chinese population=e-1.163+0.033Age+1.032DRE-2.821LogePV+2.292LogPSA1+e-1.163+0.033Age+1.032DRE-2.821LogPV+2.292LogPSAUsing the information from multivariate regression analysis, a nomogram model was developed that allows the calculation of an individual patient’s risk for a positive initial prostate biopsy. A concordance index of0.848, representing predictive accuracy, was found upon internal validation of the nomogram.Our nomogram was superior in predictive accuracy to that with diagnostic PSA data, increasing the area under the curve (AUC) from79.7%to84.8%.(2) Of the189men included in the study,78and111had a PSA of4.0-10.Ong/ml and10.1-20.0ng/ml, respectively. A total of40patients (21.2%) were diagnosed with prostate cancer. In78men with a PSA of4.0-10.0ng/ml,16(20.5%) were diagnosed with prostate cancer. In111men with a PSA of10.1-20.0ng/ml,24(21.6%) were diagnosed with prostate cancer. The rate of prostate cancer between men with a PSA of4.0-10.0and10.1-20.0ng/ml groups was not statistically different (P=0.854). The differences of age, prostate transition zone volume between prostate cancer and non-prostate cancer groups were both not statistically significant different (P=0.680and0.293, respectively). The TZPSAD and rate of positive digital rectal examination between prostate cancer and non-prostate cancer groups were both statistically different (P=0.004and0.026, respectively). The areas under the ROC curve (AUCs) for PSA and TZPSAD as continuous variables in predicting prostate cancer in men with a PSA of4.0-10.0ng/ml were0.569and0.702, respectively. When the best PSA cut-off of7.0ng/ml was chosen to predict prostate cancer, the total efficiency was30.8%and corresponding specificity and sensitivity were56.2%and54.8%, respectively. For the best TZPSAD cut-off of0.370ng/ml/ml, the total efficiency was49.9%and the corresponding specificity and sensitivity were68.8%and72.6%, respectively. If PSA7.0ng/ml was chosen for the biopsy cut-off,56.2%(9/16) of the prostate cancer could be diagnosed and the positive biopsy rate was24.3%(9/37). However, if TZPSAD0.370ng/ml/ml was chosen for the biopsy cut-off,68.8%(11/16) of the prostate cancer could be diagnosed and the positive rate of prostate biopsy was up to39.3%(11/28). At least9men could avoid unnecessary prostate biopsy. The AUCs for PSA and TZPSAD as continuous variables in predicting prostate cancer in men with a PSA of10.1-20.0ng/ml were0.463and0.730, respectively. When the best PSA cut-off of13.3ng/ml was chosen to predict prostate cancer, the total efficiency was23.7%and corresponding specificity and sensitivity were66.7%and35.6%, respectively. For the best TZPSAD cut-off of0.500ng/ml/ml, the total efficiency was49.6%and the corresponding specificity and sensitivity were70.8%and70.1%, respectively. If PSA13.3ng/ml was chosen for the biopsy cut-off,66.7%(16/24) of the prostate cancer could be diagnosed and the positive biopsy rate was21.9%(16/73). But, if TZPSAD0.500ng/ml/ml was chosen for the biopsy cut-off,70.8%(17/24) of the prostate cancer could be diagnosed and the positive rate of prostate biopsy was up to39.7%(17/43). At least30men could avoid unnecessary prostate biopsy.(3) A total of4206men without prostate cancer were included. The median initial PSA value in these men was0.6ng/ml. Of these men,1026(24.4%),177(4.2%), and90(2.1%) had a initial PSA of≥1.0≥2.5, and≥4.0ng/ml, respectively. A total of417men had their PSA measurement twice or more. The median PSA velocity in these men was0.03ng/ml/year. Of these men,25(6.0%),13(3.1%), and8(1.9%) had a PSA velocity of≥0.35、≥0.75、≥2.00ng-ml/year,respectively. These were no direct correlations between initial PSA age and initial PSA, initial PSA age and PSA velocity, and initial PSA and PSA velocity (correlation coefficient r=0.077,-0.011and-0.008, respectively; P=0.115,0.831,and0.875, respectively). After a follow-up of up to7.1years from baseline PSA measurement, the risk of PSA≥2.5ng/ml, stratified by median initial PSA (0.6ng/ml) was significantly different (log-rank test, P<0.001).(4) Of the261subjects included in the study,95(36.4%) were diagnosed with prostate cancer. The median and mean PVs in our study cohort were60and68ml, respectively. Median and mean diagnostic PSA in our study cohort were19.3and22.3ng/ml, respectively. The median age, PSA, PV, and rate of positive DRE in men diagnosed with prostate cancer were all statistically significantly higher than those did not have prostate cancer (all p values<0.05). On univariate analysis, the risk of prostate cancer was inversely associated with PV. On stepwise multivariate analysis, PV (as either a continuous or categorical variable), age, PSA, and DRE status were all found to be independent predictors of prostate cancer. Men with a PV of≥60ml (median) were found to be at a decreased risk of having prostate cancer with odds ratios of0.23when compared to those with a PV of <60ml (p value<0.010). The rates of prostate cancer in men with a PV of<60and≥60ml in the PSA of10-19.9ng/ml group were40.6%and15.1%, respectively, while rates for those with a PSA of20-50ng/ml were65.1%and26.8%, respectively. The rates of prostate cancer in for men with a PV of<60or≥60ml and with a positive DRE were71.9%and35.3%, respectively, while the rates for those with a negative DRE saw rates of41.9%and16.7%, respectively.(5) Of the251men included in the study,101and140had a PSA of4.0-10.0ng/ml and10.1-20.0ng/ml, respectively. A total of58patients (23.1%) were diagnosed with prostate cancer. In101men with a PSA of4.0-10.0ng/ml,18(17.8%) were diagnosed with prostate cancer. In140men with a PSA of10.1-20.0ng/ml,40(26.7%) were diagnosed with prostate cancer. The rate of prostate cancer between men with a PSA of4.0-10.0and10.1-20.0ng/ml groups was not statistically different (P=0.103). The differences of f/t PSA and prostate volume between prostate cancer and non-prostate cancer groups were both statistically significant different (all P values<0.05). The PSA between prostate cancer and non-prostate cancer groups were both not statistically different (P=0.361and0.435,respectively).The rates of prostate cancer inversely decreased with f/t PSA increased. Prostate cancer detection rates in men with f/t PSA<10%in PSA4.0-10.0and10.1-20.0ng/ml ranges were46.7%and40.0%, respectively. Whereas, prostate cancer detection rates in men with f/t PSA>25%in PSA4.0-10.0and10.1-20.0ng/ml ranges were only11.1%and11.8%, respectively. The AUCs for PSA and f/t PSA as continuous variables in predicting prostate cancer in men with a PSA of4.0-10.0ng/ml were0.569and0.305, respectively. When the best PSA cut-off of8.2ng/ml was chosen to predict prostate cancer, the total efficiency was33.2%and corresponding sensitivity and specificity were50.0%and66.3%, respectively. For the best f/t PSA cut-off of18%, the total efficiency was15.4%and the corresponding sensitivity and specificity were27.8%and55.4%, respectively. The AUCs for PSA and f/t PSA as continuous variables in predicting prostate cancer in men with a PSA of10.1-20.0ng/ml were0.458and0.358, respectively. When the best PSA cut-off of15.5ng/ml was chosen to predict prostate cancer, the total efficiency was23.3%and corresponding sensitivity and specificity were40.0%and58.2%, respectively. For the best f/t PSA cut-off of12%, the total efficiency was19.1%and the corresponding sensitivity and specificity were52.5%and36.4%, respectively. The AUCs for PSA and f/t PSA as continuous variables in predicting prostate cancer in men with a PSA of4.0-20.0ng/ml were0.553and0.325, respectively. When the best PSA cut-off of10.5ng/ml was chosen to predict prostate cancer, the total efficiency was29.9%and corresponding sensitivity and specificity were62.1%and48.2%, respectively. For the best f/t PSA cut-off of12%, the total efficiency was15.0%and the corresponding sensitivity and specificity were53.4%and28.0%, respectively. In men with PSA4.0-10.0and10.1-20.0ng/ml, the efficiencies of best cutoffs for f/t PSA were not better than best cutoffs for PSA in diagnosis of prostate cancer.CONCLUSION (1) Our results indicate that the risk of a positive initial prostate biopsy can be predicted to a satisfactory level in a Chinese population using our nomogram. The nomogram can be used to identify and counsel patients who should consider a prostate biopsy, ultimately enhancing accuracy in diagnosing prostate cancer.(2) Using TZPSAD can improve the efficiency of PSA in diagnosis of prostate cancer and decreases the unnecessary prostatic biopsy in men with a PSA of both 4.0-10.0and10.1-20.0ng/ml.(3) The median baseline PSA and PSA velocity in men younger than50years old with prostate cancer are0.6ng/ml and0.03ng/ml/year, respectively. Men younger than50years old without prostate cancer with an initial PSA higher than median (0.6ng/ml) have a subsequently higher risk of PSA value≥2.5ng/ml.(4) PV is an independent predictor of prostate cancer in men with a PSA of10-50ng/ml. In clinical practice, especially for those countries with lower PCa incidence, PV should be taken into account so as to improve the prostate cancer detection rate and avoid unnecessary biopsy.(5) f/t PSA does not improve the efficiency of prostate specific antigen in diagnosis of prostate cancer in Chinese men with PSA4.0-10.0and10.1-20.0ng/ml. |
PDF Full Text Request