- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
Despite some controversy, it has previously been reported that PSA may not perform as accurately as PSA density (PSAD) in predicting outcomes after radical prostatectomy (RP) among patients with Gleason score (GS) 6 prostate cancer (PCa) . Initially introduced to enhance the detection of PCa, PSAD has also been shown to be a useful predictor of adverse pathological features and biochemical outcome in patients undergoing RP [2–5]. Some have reported that PSAD is a significant independent predictor of GS upgrading after RP, even when accounting for PSA . However, in reality, there are few studies comparing the usefulness of PSA with that of PSAD in predicting upgrading after RP. Also, most published studies on the prediction of upgrading included significant proportions of subjects who did not undergo contemporary multicore prostate biopsy. Moreover, most studies from major academic centres on the potential usefulness of PSAD as a preoperative predictor of pathological and/or biochemical outcomes after RP have not included detailed biopsy core-related data, such as percentage of positive cores or maximum tumour length in a biopsy core, in their analyses, citing that most of their subjects were referred for surgery after undergoing prostate biopsy elsewhere. Since biopsy-related factors have also been known to be useful prognosticators in patients undergoing RP [7,8], it can be suggested that objective comparison of PSA and PSAD as a preoperative predictor of Gleason score upgrading after RP in contemporary patients should also incorporate the effects of variables obtained from prostate biopsies done using a contemporary approach. Thus, we performed a study analysing our prospectively collected database to compare the respective merits of PSAD and PSA in the prediction of GS upgrading after RP in men who were clinically diagnosed with Gleason 6 PCa using contemporary multicore (≥12 cores) prostate biopsy and who subsequently also underwent RP at our institution.
PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
After obtaining institutional review board approval, we analysed the prospectively collected database of 1204 patients who had undergone RP at our institution from November 2003 to April 2010. Among the 1204 patients, 269 patients who did not receive prostate biopsy at our institution were excluded from our analysis. Also excluded were men who had received prostatic surgery (n= 49) or neoadjuvant treatment (n= 20) preoperatively and those with missing data (n= 21). After such exclusions, a total of 845 patients remained. Of the 845, we selected 505 patients who were diagnosed with biopsy GS 6 PCa from prostate biopsy at our institution and these 505 patients were included in our study. At our institution, all TRUS-guided prostate biopsies were performed using a multicore scheme during the aforementioned period with relevant detailed biopsy core data collected prospectively. Before obtaining biopsy cores, prostate volume was routinely measured using TRUS. TRUS volumes were assessed using the prostate ellipsoid formula. In all cases, prostate was routinely biopsied near the base, mid-gland and apex, bilaterally, with at least six biopsies per side; thus, 12 baseline biopsy cores were taken in all men, and additional biopsies were taken to include suspicious appearing lesions if needed.
Data assessed included patient age, body mass index (BMI), PSA level, biopsy GS, total number of biopsy cores obtained, percentage of positive cores, maximum tumour length in a core, percentage of total tumour length (total tumour length/total core length), clinical stage, TRUS volume, PSAD (PSA divided by TRUS volume), pathological Gleason grade, pathological stage and marginal status. Pathological analyses of both biopsy and RP specimens were uniformly performed by a single genitourinary pathologist with more than 20 years of experience.
The SPSS software package version 15.0 (Statistical Package for Social SciencesTM, Chicago, IL, USA) was used for statistical analysis. For our analyses, age and TRUS volumes were considered as continuous variables. PSA and PSAD were evaluated as continuous and categorical variables. BMI was treated as a dichotomous variable according to the Asia-Pacific definition of obesity (<25 kg/m2 vs ≥25 kg/m2) since there were too few men with BMI ≥30 kg/m2. Logistic regression was used to predict high (≥7) pathological GS. When comparing patients with and without GS upgrading, we assessed the difference in clinicopathological profile of patients using the chi-squared test, Fisher exact test, and Mann–Whitney test. Multivariate analyses were performed according to logistic regression models to identify independent predictors of GS upgrading. Variables associated with biopsy cancer volume were included one at a time for each multivariate model, as they were not independent of each other. A receiver operator characteristic (ROC) curve was used to analyse the predictive performances of the multivariate model. Area under the curves (AUCs) of ROC curves were compared using the Mantel–Haenszel test. Statistical significance was defined as a two-tailed P < 0.05.
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
For the 505 total patients who all had biopsy GS 6, the mean (sd) patient age was 65.2 (6.8) years, mean (sd) BMI 24.3 (2.5) kg/m2, mean (sd) TRUS volume 40.5 (17.6) mL, and mean PSA 7.8 (4.1) ng/mL. Overall, mean (sd) PSAD was 0.21 (0.16) ng/mL2. Of the 505 patients, 340 (67.3%) had 12 cores of tissue extracted at biopsy and the other 165 (32.7%) patients had ≥13 cores extracted. Upgrading to pathological GS 7 was found in 253 patients (50.1%), and to pathological GS 8 in three patients (0.6%). No patient had upgrading to pathological GS ≥9.
As shown in Table 1, patients with upgrading after RP were older (P= 0.008) than those without upgrading and these patients had significantly higher PSA levels (P= 0.008) and PSAD values (P < 0.001) than those without upgrading. Meanwhile, the two groups of patients showed significant differences in various factors associated with biopsy cancer volume, including number and percentage of positive biopsy cores, maximum tumour length in a core, percentage of total tumour length in biopsy cores, and maximum percentage of tumour length in a biopsy core (all P < 0.001). No significant difference was noted regarding TRUS volume (P= 0.076) and total number of biopsy cores extracted (P= 0.136) between the two groups. As for the pathological features observed in RP specimens, patients with upgrading had significantly higher rates of extracapsular extension and positive surgical margins (all P < 0.001).
Table 1. Clinicopathological characteristics of patients with and without GS upgrading after RP
|Variable||Total||Pathological GS|| P |
|Age, years|| || || ||0.008|
| Mean (sd)||65.20 (6.78)||64.31 (7.14)||65.93(6.38)|| |
| Median (range)||66 (43–79)||66 (43–76)||67 (47–79)|| |
|BMI|| || || ||0.646|
| Mean (sd)||24.27 (2.53)||24.33 (2.49)||24.23 (2.58)|| |
| Median (range)||24.15 (15.62–34.01)||24.28 (17.17–31.99)||24.09 (15.62–34.01)|| |
|PSA, ng/mL|| || || ||0.008|
| Mean (sd)||7.79 (7.07)||6.90 (5.36)||8.53 (8.16)|| |
| Median (range)||5.90 (1.12–90.90)||5.17 (1.12–43.54)||6.30 (1.41–90.90)|| |
|PSAD, ng/mL2|| || || ||<0.001|
| Mean (sd)||0.21 (0.16)||0.16 (0.11)||0.24 (0.19)|| |
| Median (range)||0.16 (0.02–1.42)||0.14 (0.02–0.74)||0.18 (0.04–1.42)|| |
|TRUS volume, mL|| || || ||0.076|
| Mean (sd)||39.47 (17.55)||40.39 (19.71)||38.98 (14.77)|| |
| Median (range)||36.6 (10.8–141.)||43.0 (18.0–141.0)||41.1 (10.8–99.9)|| |
|Clinical stage, n (%)|| || || ||0.771|
| T1||362 (71.7)||181 (72.7)||181 (70.7)|| |
| T2||143 (28.3)||68 (27.3)||75 (29.3)|| |
|No. total cores sampled at biopsy, n (%)|| || || ||0.136|
| 12||340 (67.3)||161 (64.7)||179 (69.9)|| |
| ≥13||165 (32.7)||88 (35.3)||77 (30.1)|| |
|Percentage of positive cores|| || || ||<0.001|
| Mean (sd)||20.97 (16.69)||16.13 (14.81)||24.99 (17.11)|| |
| Median (range)||16.67 (5.00–100.00)||8.33 (5.00–100.00)||21.43 (6.67–85.71)|| |
|Maximum tumour length in a core|| || || ||<0.001|
| Mean (sd)||0.35 (0.29)||0.25 (0.22)||0.43 (0.32)|| |
| Median (range)||0.30 (0.02–1.60)||0.20 (0.02–1.20)||0.40 (0.02–1.60)|| |
|Percentage of total tumour length in cores|| || || ||<0.001|
| Mean (sd)||3.75 (4.86)||2.35 (4.07)||4.91 (5.15)|| |
| Median (range)||1.95 (0.08–40.76)||1.09 (0.09–40.76)||3.00 (0.08–31.88)|| |
|Maximum percentage of tumour length in a core|| || || ||<0.001|
| Mean (sd)||22.70 (18.59)||17.06 (15.76)||27.39 (19.47)|| |
| Median (range)||17.65 (1.05–100.0)||11.76 (1.05–100.0)||23.08 (1.11–93.33)|| |
|Extracapsular extension, n (%)||52 (10.3)||6 (2.4)||46 (17.9)||<0.001|
|Seminal vesicle invasion, n (%)||4 (0.8)||0||4 (1.6)||0.088|
|Positive surgical margin, n (%)||102 (20.2)||28 (11.2)||74 (28.9)||<0.001|
To identify significant predictors of upgrading after RP, uni- and multivariate analyses incorporating preoperative variables were performed. The results of univariate analyses were similar to findings shown in Table 1. When predictive accuracies were assessed for each variable using AUC assessed from ROC curves, PSAD showed significantly higher accuracy than PSA when compared one-on-one (0.648 vs 0.580, P < 0.001) (Fig. 1). In performing multivariate logistic regression analyses, aforementioned variables found to be significant in univariate analyses were included (Table 2). As shown in Table 2, various multivariate models were constructed to identify significant predictors of GS upgrading by including PSA or PSAD along with other variables. Each multivariate model incorporated a variable showing biopsy cancer volume. Both PSA and PSAD were observed to be independent predictors of upgrading in all versions of multivariate models (all P < 0.05). Among the models 1 to 4 as shown in Table 2, the predictive accuracy of the model incorporating PSAD was observed to be significantly higher than those using PSA in models 1, 2 and 4, as assessed by comparing AUCs of ROC curves (model 1, P= 0.048; model 2, P= 0.002; model 3, P= 0.201; model 4, P= 0.044).
Table 2. Multivariate logistic regression models for the prediction of upgrading after RP
|Predictor||Multivariate analysis with PSA||Multivariate analysis with PSAD|
| P ||OR (95% CI)|| P ||OR (95% CI)|
|Model 1|| || |
| Age||0.004||1.043 (1.014–1.073)||0.001||1.052 (1.022–1.083)|
| PSA||0.012||1.506 (1.093–2.074)|| ||not applicable|
| PSAD|| ||not applicable||<0.001||2.247 (1.603–3.150)|
| Percentage of positive cores||<0.001||1.041 (1.027–1.056)||<0.001||1.035 (1.020–1.050)|
|Model 2|| || |
| Age||0.017||1.035 (1.006–1.064)||0.003||1.044 (1.014–1.075)|
| PSA||0.015||1.492 (1.080–2.062)|| ||not applicable|
| PSAD|| ||not applicable||<0.001||2.253 (1.603–3.168)|
| Max. tumour length in a core||<0.001||3.531 (2.376–5.246)||<0.001||2.975 (1.981–4.466)|
|Model 3|| |
| Age||0.007||1.040 (1.011–1.070)||0.002||1.047 (1.018–1.078)|
| PSA||0.043||1.397 (1.011–1.929)|| ||not applicable|
| PSAD|| ||not applicable||<0.001||2.080 (1.474–2.935)|
| Percentage of total tumour length in biopsy cores||<0.001||1.177 (1.109–1.249)||<0.001||1.142 (1.075–1.213)|
|Model 4|| |
| Age||0.006||1.040 (1.011–1.070)||0.001||1.048 (1.018–1.079)|
| PSA||0.036||1.414 (1.023–1.954)|| ||not applicable|
| PSAD|| ||not applicable||<0.001||2.123 (1.505–2.994)|
| Max. percentage of tumour length in a core||<0.001||1.036 (1.024–1.049)||<0.001||1.030 (1.017–1.043)|
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
In the present study, PSAD, when compared with PSA, was observed to be a significantly more accurate predictor of GS upgrading after RP among patients diagnosed with biopsy Gleason 6 PCa. According to this finding, PSAD may indeed be a valuable tool in identifying patients actually harbouring more aggressive disease among those seeking less invasive treatment. It should be recalled that our findings were obtained from patients who all underwent contemporary multicore prostate biopsy using the same scheme at our institution. Furthermore, it should be noted that the superiority of PSAD over PSA as a predictor of upgrading was confirmed using multivariate analyses incorporating detailed biopsy core data obtained from a prospectively collected database.
Previously, others have reported on the potential predictors of GS upgrading after RP [10–13], but few studies have specifically compared the usefulness of PSA and PSAD in the prediction of GS upgrading. Analysing patients who underwent RP for a single microfocal Gleason 6 PCa detected from biopsy, Thong et al.  reported that age >65 years and PSAD >0.20 ng/mL/g, but not PSA, were independent predictors of adverse pathological features, including GS upgrading. Also, Magheli et al.  reported that PSAD was a significant independent predictor of upgrading even when accounting for PSA in patients with Gleason 6 PCa. In their study, a comparison of the accuracy of multivariate logistic regression models, devised to predict upgrading, also found significantly higher accuracy for the multivariate model using PSAD than for the model using PSA. In Thong et al.'s study , PSAD was calculated by dividing PSA by weight of RP specimen rather than TRUS volume. In addition, significant proportions of subjects included in the aforementioned studies had <12 cores extracted during prostate biopsy [6,10–14] and detailed data on biopsy cancer volume were not incorporated into multivariate analyses in these studies. Despite some differences in the study design and methodology, the results of these two studies can be considered to be supportive of our findings.
Potential explanations for the observed superiority of PSAD over PSA in the prediction of GS upgrading could lie in the significance of prostate volume. Approaching statistical significance, TRUS volume was smaller in upgraded patients compared with those without upgrading in our study. Freedland et al.  reported that men with a small prostate were found to have more advanced disease and be at greater risk for progression after RP than those with larger glands. Also, from analysing only the patients who underwent multisite extended prostate biopsy, Kassouf et al.  reported that patients with a large prostate volume have a significantly higher incidence of well differentiated tumour at RP and a lower risk of GS upgrading. Whether the observed association between prostate size and GS is attributable to a certain biological mechanism, or PSA-driven biopsies in patients with larger glands, remains to be elucidated. However, as prostate volume has generally been shown to be associated with GS, PSAD would probably prove more informative than PSA with regard to GS upgrading.
Previous studies on the comparison of PSA and PSAD as predictors of various outcomes, other than GS upgrading, after RP have shown inconsistent findings. Jones et al.  have reported that PSAD did not outperform PSA for predicting adverse pathological outcomes and biochemical recurrence after RP. Others have found that PSA was significantly better than PSAD in predicting tumour volume and biochemical recurrence after RP using ROC analysis . By contrast, in stratifying patients according to GS, some have observed that PSAD was significantly better than PSA alone for predicting extraprostatic extension and biochemical-free recurrence in patients with a biopsy GS ≤6 . Obviously, various factors, including differences in biopsy scheme, may have played a role in the observed differences in the results of the aforementioned studies. Furthermore, it should be noted that detailed biopsy core data were not incorporated into the analyses in most of these studies. In the present study, parameters of biopsy cancer volumes were all shown to be significant predictors of upgrading on multivariate analyses. Such a finding would indicate that additional information on biopsy cancer volume may well have made differences regarding the results of aforementioned studies. In addition, PSAD was calculated using different methods in the aforementioned studies. Some used TRUS volume while others applied volume or weight of RP specimen in defining PSAD. Interobserver variability would certainly influence the accuracy of TRUS as well. However, since only TRUS volume would be available preoperatively, PSAD can only be calculated using TRUS volume for the purpose of applications in the preoperative counselling of patients, as done in our study.
As can be seen in the literature, active surveillance is gaining acceptance as a primary treatment in patients diagnosed with low-risk PCa, which is defined by PSA level, clinical stage, and biopsy Gleason sum . However, as widely acknowledged, a significant proportion of the low-risk PCa diagnosed currently is found to have more adverse pathological features after RP. Accordingly, efforts are being directed to accurately select men harbouring more aggressive disease among those who are initially diagnosed with low-risk PCa so that such men can be offered curative primary treatment. According to our findings, the assessment of prostate volume to obtain PSAD would prove clinically significant, especially in patients with low-risk PCa who are considering active surveillance as a primary treatment.
One of potential limitations of the present study may be the fact that the study population was limited to patients who underwent both prostate biopsy and RP at our centre. Patients who opted for radiation treatment or other primary treatment methods were not included. Also, the upgrading rate observed among our subjects may be considered relatively high for a contemporary RP cohort. It should be noted that there has been a change in GS grading in that pathologists are now more likely to assign worse GS than they would have in the past . However, as intervals from biopsy to RP were generally not longer than 3 months among our subjects, GS grading of biopsy and RP specimen may well have been performed in similar fashion in each patient. Lastly, although significant differences were observed between PSA and PSAD AUCs, the actual clinical impact of such differences remains to be determined.
To conclude, according to the present study with prospectively collected data, PSAD may be significantly more accurate in the prediction of GS upgrading after RP than PSA, even when accounting for detailed parameters of biopsy cancer volume in the current era of extended prostate biopsies. This finding would support the inclusion of PSAD into the risk stratification system for PCa patients seeking less invasive treatment, such as active surveillance. Further investigation, preferably via a prospective study, would be needed to evaluate the true role of PSAD in the prediction of various outcomes after primary therapy for PCa in the contemporary era.