Fax: (011) 49 30450515904
The ratio of prostate-specific antigen (PSA) to prostate volume (PSA density) as a parameter to improve the detection of prostate carcinoma in PSA values in the range of < 4 ng/mL
Article first published online: 8 JUL 2005
Copyright © 2005 American Cancer Society
Volume 104, Issue 5, pages 993–1003, 1 September 2005
How to Cite
Stephan, C., Stroebel, G., Heinau, M., Lenz, A., Roemer, A., Lein, M., Schnorr, D., Loening, S. A. and Jung, K. (2005), The ratio of prostate-specific antigen (PSA) to prostate volume (PSA density) as a parameter to improve the detection of prostate carcinoma in PSA values in the range of < 4 ng/mL. Cancer, 104: 993–1003. doi: 10.1002/cncr.21267
- Issue published online: 17 AUG 2005
- Article first published online: 8 JUL 2005
- Manuscript Accepted: 20 APR 2005
- Manuscript Revised: 13 APR 2005
- Manuscript Received: 16 DEC 2004
- area under the receiver operating characteristic curve;
- prostate carcinoma;
- prostate-specific antigen;
- prostate-specific antigen density
The objective of this study was to evaluate the prostate specific antigen (PSA) density (PSAD) (the quotient of PSA and prostate volume) compared with the percent free PSA (%fPSA) in different total PSA (tPSA) ranges from 2 ng/mL to 20 ng/mL. Possible cut-off levels depending on the tPSA should be established.
In total, 1809 men with no pretreatment of the prostate were enrolled between 1996 and 2004. Total and free PSA were measured with the IMMULITE PSA and Free PSA kits (Diagnostic Products, Los Angeles, CA). Prostate volume was determined by transrectal ultrasound. The diagnostic validity of tPSA, %fPSA, and PSAD was evaluated by receiver operation characteristic (ROC) curve analysis.
The PSAD differed significantly (P < 0.0001) between patients with prostate carcinoma and patients with benign prostatic hyperplasia in all analyzed ranges of tPSA and prostate volume. At the 90% and 95% sensitivity levels and regarding the area under the ROC curve (AUC) within the tPSA range of 2–4 ng/mL, The PSAD was significantly better than tPSA and %fPSA. Within the tPSA range of 4–10 ng/mL, the PSAD did not perform better than %fPSA.
PSAD showed a better performance than %fPSA at tPSA concentrations < 4 ng/mL for detecting prostate carcinoma, with a significantly larger AUC for PSAD (0.739) compared with %fPSA (0.667). PSAD did not perform better than %fPSA when the tPSA range of 4–10 ng/mL was analyzed. Different PSAD cut-off values of 0.05 at tPSA 2–4 ng/mL, 0.1 at tPSA 4–10 ng/mL, and 0.19 at 10–20 ng/mL were necessary to reach 95% sensitivity. Cancer 2005. © 2005 American Cancer Society.
Prostate carcinoma (PCa) is one of the most common malignancies among men in the Western world. Increasing mortality rates due to PCa have been observed worldwide. According to the World Health Organization, PCa caused an estimated 239,000 deaths in 1998 and an estimated 269,000 deaths in 2002 (www.who.org/whr). This disease usually progresses imperceptibly; thus, patients are unlikely to seek medical help during the early stages. For these reasons, screening programs aimed at early detection have been developed.1 The prostate-specific antigen (PSA) test is among the best screening tools available in medicine today and is recognized as the best marker for its early detection.2 However, the lack of specificity is an important limitation of PSA. Increased PSA concentrations are found not only in patients with PCa but also in patients with benign prostatic diseases.3
Various approaches were developed to improve the low specificity of total PSA (tPSA), which circulates in the serum bound to α1-antichymotrypsin (ACT) as PSA-ACT and as free PSA (fPSA).4, 5 The recognition that PSA exists in a variety of molecular forms has resulted in a great opportunity for improving the specificity of PSA testing. Many investigators have demonstrated that the fPSA/tPSA ratio (the percent free PSA [%fPSA]) results in a significant enhancement in specificity.4–9 Among patients with elevated tPSA, men with benign prostate hyperplasia (BPH) tend to have higher %fPSA ranges than men with PCa.10
Despite these encouraging results, several problems remain with regard to the %fPSA. It has been confirmed that %fPSA increases with older age9 and prostate size.11, 12 In addition, using the %fPSA to discriminate between PCa and BPH has yielded significant results only in patients with small prostates but not in patients with larger glands.11, 12 There is no fixed cut-off size for differentiating small and large glands. Some studies used 30–35 mL as a cut-off size, and others used 40 mL or 60 mL.9, 11–13
Other calculated parameters, such as PSA velocity as PSA change over time14 or age-specific PSA reference ranges,15 also were introduced to enhance the specificity of tPSA. However, an advantage compared with %fPSA could not be observed.16–18 In the early 90s Benson et al.19, 20 first introduced the term PSA density to correct tPSA for prostate volume, because PCa releases more PSA per unit volume into the circulation than BPH.21 Thus, it is suspected that the PSA density (PSAD), as a quotient of tPSA to prostate volume, is higher in patients with PCa than in patients with BPH. Cut-off levels from 0.1 ng/mL/cm3 to 0.2 ng/mL/cm3 are used to differentiate the two diseases.
Many studies (with > 100 patients included) have been enrolled to investigate the effectiveness of PSAD compared with PSA and %fPSA.16, 17, 22–32 Although most studies showed an advantage of using PSAD compared with PSA alone, only a few studies could demonstrate further a better performance of PSAD compared with %fPSA.32–34 Two of those studies included relatively small numbers of patients (n = 43 patients and n = 105 patients).33, 34 The other, very recent study by Sakai et al.32 in 400 Japanese men demonstrated a better performance of PSAD and, surprisingly, tPSA compared with %fPSA. This is an opposite finding compared with many other studies.4–9, 35 To date, only 1 study has been performed in > 1000 men to verify the parameters PSA, PSAD, and %fPSA.29 In 1913 Japanese men who had a tPSA from 2 ng/mL to 20 ng/mL, the specificity of all parameters was low for men who had prostate volumes > 40 mL but was better in men who had smaller glands (< 20 mL).29 The authors of that report concluded that reference values for tPSA, %fPSA, and PSAD must be changed according to prostate volume.29 Although the Japanese study had a large sample size, there are differences in PCa detection rates and other PCa parameters between Japanese men and Caucasian men, making their application on the Western world difficult. The objectives of the current study were 1) to evaluate PSAD specifically compared with %fPSA in different tPSA ranges from 2 ng/mL to 20 ng/mL, 2) to evaluate PSAD within different prostate size ranges in a large Caucasian population, and 3) to establish possible cut-off levels depending on the tPSA.
MATERIALS AND METHODS
In total, 1809 men (ages 38–89 years) with no pretreatment of the prostate and with tPSA in the range of 2–20 ng/mL were enrolled in the study between March, 1996 and March, 2004. The distribution of all patients with PCa (n = 1148 men) and BPH (n = 661 men) within all analyzed tPSA ranges is shown in Table 1.
|Variable||Total PSA range|
|2–4 ng/mL||4–10 ng/mL||10–20 ng/mL||2–10 ng/mL||2–20 ng/mL|
|No. of patients||196||111||348||627||117||410||544||738||661||1148|
|%Free PSA (%)|
|Prostate volume (mL)|
|Suspect DRE (%)||4.95||63.3||13.2||63||13.7||70.2||10.4||63||11||65.6|
|PSA density (ng/mL/cm3)|
All men were seen in the Department of Urology or in the affiliated outpatient department at the University Hospital Charité (Berlin, Germany). Indications for referral were PSA elevations, BPH symptoms, abnormal digital rectal examination (DRE), or biopsy-proven PCa, which explains the high number of patients with PCa. Therefore, the current study population does not represent a typical urologically referred population or a screening population.
All serum samples were drawn before any prostate manipulation (or at least 3–4 weeks after an earlier manipulation) and centrifuged within 2–3 hours after sampling. Samples were analyzed immediately or were stored at − 20 °C for no longer than 48 hours before they were assayed. The IMMULITE PSA and Free PSA kits (Diagnostic Products, Los Angeles, CA) were used to measure tPSA and fPSA, respectively. Assays were solid-phase, 2-site, sequential chemiluminescent immunometric tests, which are performed automatically on the IMMULITE automated analyzer with analytic sensitivities of 0.02 ng/mL and 0.03 ng/mL for fPSA and tPSA, respectively. The tests use both polyclonal antibodies and monoclonal antibodies that are specific for PSA or an anti-PSA monoclonal antibody that is specific only for fPSA.
Prostate volume was determined by TRUS (Combison 330; Kretz Technik, Zipf, Austria) using the prolate ellipse formula (height × width × length × 0.52). Only well trained urologists examined the patients. The interobserver and intraobserver variation of 13–15% did not differ from that found in earlier investigations.35 All men underwent DRE. A DRE finding that was nonsuspicious for carcinoma was defined as negative, and a finding that was suspicious for carcinoma was considered positive.
All patients had their disease confirmed histologically. All patients with PCa had their disease confirmed by sextant or octant biopsy. The histologic diagnosis of the patients without PCa (the BPH group) was confirmed on the basis of tissue obtained by transurethral resection of the prostate or open adenomectomy (n = 114 patients), by TRUS-guided sextant or octant biopsy (n = 547 patients), or by both (n = 65 patients). The diagnosis was BPH with or without prostatitis or with no histologic abnormality but with no evidence of prostatic intraepithelial neoplasia. Differentiation based on the number of biopsy sessions and biopsy history was not performed.
Conventional statistical calculations were performed using the statistical software package SPSS 11.5 for Windows (SPSS, Chicago, IL). The Mann–Whitney U test, the nonparametric Kruskal–Wallis test of variance, and the rank correlation coefficients according to Spearman (rs) were calculated. The diagnostic validity of tPSA, %fPSA, and PSAD was evaluated by receiver operation characteristic (ROC) curve analysis. The software package GraphROC 2.1 for Windows was used to calculate the area under the curve (AUC).36 Significance was defined as P < 0.05.
The PSAD (the ratio of tPSA to prostate volume) differed significantly (P < 0.0001) between PCa and BPH in all analyzed tPSA ranges (Table 1). A clear trend toward higher PSAD values with increasing tPSA concentrations was observed. The subdivision into 3 groups with increasing prostate volume (Table 2) also revealed the significant differences in all 3 groups (P < 0.0001) between PCa and BPH regarding PSAD. However, the largest difference between the median PSAD values was visible in patients with small glands (0.15 vs. 0.3). In addition, for both patients with BPH and patients with PCa, PSAD differed significantly between men with abnormal DRE status and men with nonsuspicious DRE status (P < 0.0001).
|Variable||Prostate volume range|
|Volume < 40 mL||Volume 40–60 mL||Volume > 60 mL|
|No. of patients||348||872||177||200||179||76|
|Prostate volume (mL)|
|Suspect DRE (%)||13.5||69.3||6.2||58||10.6||43.4|
|PSA density (ng/mL/cm3)|
Correlations of tPSA, %fPSA, Age, Prostate Volume, and DRE Status
The distribution of the 1809 patients is shown in Table 1. In all tPSA ranges, the tPSA was significantly lower in the BPH group (P < 0.0001) compared with the PCa group except among men who had tPSA in the 10–20 ng/mL range (P = 0.438). The %fPSA was significantly lower (P < 0.0001) in patients with PCa for all analyzed tPSA ranges. A negative correlation between tPSA and %fPSA is illustrated in Table 1 (rs, − 0.335; P < 0.0001 for all patients). The %fPSA values decreased with increasing tPSA concentrations. In the 2–4 ng/mL, 4–10 ng/mL, and 10–20 ng/mL tPSA ranges, the median %fPSA was 16.4%, 14.5%, and 13%, respectively, in the BPH group and 11.4%, 9.7%, and 7.4% in the PCa group, respectively.
The median age for the patients with PCa was 64 years (range, 40–86 years), and it was 66 years (range, 38–89 years) for the patients with BPH. In all investigated groups, with exception of men with tPSA in the 2–4 ng/mL range (P = 0.9), age was statistically different between the PCa group and the BPH group (P values from 0.01 to < 0.0001). Age was correlated positively with %fPSA in both the PCa group and the BPH group in all tPSA ranges (PCa: rs, 0.26–0.38; BPH: rs, 0.21–0.35), except for men in the BPH group with tPSA in the 10–20 ng/mL range (rs, 0.206; P = 0.054). There was a positive correlation of age with tPSA in the BPH group (rs, 0.12; P = 0.002) but not in the PCa group (rs, 0.04; P = 0.62).
The median prostate volume for the BPH group was significantly larger than in the PCa group within all tPSA ranges (P < 0.0001). Prostate volume was correlated positively with tPSA for both the BPH group (rs, 0.26; P < 0.0001) and the PCa group (rs, 0.1; P = 0.001). There also was a positive correlation in all tPSA ranges between prostate volume and %fPSA for the PCa group (rs, 0.28–0.33; P < 0.0001) and the BPH group (rs, 0.31–0.42; P < 0.0001).
Table 2 shows that a subdivision into gland volume ranges of < 40 mL, 40–60 mL, and > 60 mL indicates an increasing trend for the median %fPSA with higher gland volume (BPH: 13.4%, 16.4%, and 17.9%, respectively; PCa: 8.5%, 10.5%, and 13.5%, respectively). However, comparing the 3 BPH groups with one another, the difference in %fPSA was highly significant (P < 0.0001 for < 40 mL vs. 40–60 mL and < 40 mL vs. > 60 mL) but had borderline significance when comparing the groups with prostate volumes of 40–60 mL and > 60 mL (P = 0.044). The same significance levels were observed for the patients with PCa (P < 0.0001 for < 40 mL vs. 40–60 mL and < 40 mL vs. > 60 mL; P = 0.001 for 40–60 mL and > 60 mL).
The number of suspicious DRE results decreased significantly with increasing prostate volume in the patients with PCa (from P = 0.03 to P < 0.0001). For the patients with BPH, there was a significant difference only between the groups with prostate volumes of < 40 mL and 40–60 mL (P = 0.012), but the other comparisons revealed no differences in the number of suspicious DRE results (P = 0.14 and P = 0.34, respectively). The %fPSA differed between men with PCa and BPH who had abnormal DRE results (9% vs. 13.5%; P < 0.0001; n = 851 men) and who had nonsuspicious DRE results (9.4% vs. 15.3%; P < 0.0001). However, comparing %fPSA within the BPH group between men with abnormal DRE results and nonsuspicious DRE results, there was no difference (P = 0.145). In the PCa group, there also was no difference (P = 0.1).
ROC Analyses at Cut-Off Levels and AUC Comparisons
For all tPSA ranges, specificity levels at 90% with 95% sensitivity (Table 3) and sensitivity levels at 90% with 95% specificity (Table 4) for tPSA, %fPSA, and PSAD were calculated. Clearly, like what was seen previously for the median values, there was a visible, downward trend of %fPSA with higher tPSA values, especially for the 90% or 95% sensitivity cut-off levels in men who had a PSA range of 2–4 ng/mL compared men who had a PSA range of 4–10 ng/mL, in whom the cut-off difference was > 8% or > 10% (25.3% vs. 17% and 31.5% vs. 21.2%, respectively) (Table 4). PSAD showed an upward trend with higher tPSA values at all sensitivity cut-off levels (Table 4). The important 95% sensitivity cut-off level for the PSAD increased from 0.05, to 0.1, and to 0.19 for tPSA ranges 2–4 ng/mL, 4–10 ng/mL, and 10–20 ng/mL, respectively.
|Total PSA range||Specificity (%)||Sensitivity (%)||Cut-off value|
|2–4 ng/mL||90||23 (17–31)a||23 (17–31)||34 (26–42)||3.67||8.07||0.13|
|95||7 (4–13)a||15 (9–22)b||23 (17–31)||3.88||6.7||0.14|
|2–10 ng/mL||90||18 (16–20)a||31 (28–34)cd||38 (35–41)||8.56||7.87||0.24|
|95||11 (9–13)a||18 (16–21)de||26 (24–29)||9.11||6.31||0.28|
|4–10 ng/mL||90||16 (14–19)a||33 (29–36)d||35 (32–37)||8.9||7.87||0.27|
|95||5 (4–7)a||18 (16–21)cd||24 (21–27)||9.59||6.05||0.31|
|2–20 ng/mL||90||23 (21–25)a||38 (36–41)d||43 (40–45)||12.29||7.74||0.29|
|95||9 (8–10)a||24 (22–26)d||26 (24–28)||15.65||6.25||0.38|
|10–20 ng/mL||90||10 (8–13)a||48 (44–52)d||40 (36–44)||17.84||7.1||0.47|
|95||3 (2–4)a||31 (27–35)ad||19 (16–22)||19.11||5.92||0.63|
|tPSA range||Sensitivity (%)||Specificity (%)||Cut-off values|
|2–4 ng/mL||90||26 (21–32)b||16 (12–22)b||35 (29–41)||2.39||25.26||0.06|
|95||18 (14–24)a||7 (4–11)a||21 (17–27)||2.28||31.5||0.05|
|2–10 ng/mL||90||29 (26–32)c||35 (32–39)||41 (38–45)||3.57||18.28||0.1|
|95||18 (16–22)c||21 (18–24)a||29 (26–32)||2.9||22.19||0.08|
|4–10 ng/mL||90||18 (15–22)c||39 (34–43)d||36 (32–41)||4.79||16.98||0.12|
|95||11 (9–14)c||20 (17–24)e||25 (21–29)||4.41||21.15||0.1|
|2–20 ng/mL||90||30 (27–33)c||37 (34–40)f||44 (41–47)||4.09||17.32||0.12|
|95||20 (17–22)c||22 (19–25)a||28 (25–31)||3.16||21.43||0.09|
|10–20 ng/mL||90||12 (7–18)c||38 (31–46)d||47 (39–55)||10.46||14.83||0.23|
|95||6 (5–7)c||25 (18–32)e||32 (25–39)||10.18||18.88||0.19|
Table 3 shows the significance differences when comparing tPSA, %fPSA, and PSAD at 90% and 95% specificity. For the patients with low tPSA in the range of 2–4 ng/mL, in whom a high specificity is necessary to avoid general biopsies, PSAD performed significantly better than tPSA (P < 0.0001) but not significantly better than %fPSA (23% vs. 34%; P = 0.104). Conversely, there was a significant advantage for PSAD compared with %fPSA at 95% specificity for the men with tPSA ranges of 2–10 ng/mL (P = 0.003) and 4–10 ng/mL (P = 0.026); whereas, for the men with tPSA in the range of 10–20 ng/mL, %fPSA performed better than PSAD (P < 0.0001). However, at tPSA ranges up to 10 ng/mL or 20 ng/mL, only high sensitivity levels were needed.
Table 4 shows the specificity levels at given sensitivities. At the 90% and 95% sensitivity levels within the tPSA range of 2–4 ng/mL, PSAD was significantly better than tPSA and %fPSA (P = 0.0002 and P = 0.002, respectively, for 90% sensitivity; P = 0.04 and P = 0.013, respectively, for 95% sensitivity). Within the tPSA ranges of 2–10 ng/mL and 2–20 ng/mL, PSAD performed significantly better than %fPSA (P = 0.02 for each) at 95% sensitivity. In all other comparisons, PSAD was equal to %fPSA but was significantly better than tPSA. However, %fPSA could reach the significance level of tPSA in only 5 of 10 comparisons (Table 4).
ROC analysis shows that PSAD curves ran significantly above the %fPSA and tPSA curve for all tPSA ranges where data from the low tPSA (ranges: 2–4 ng/mL, 2–10 ng/mL, and 2–20 ng/mL) were included (Table 5). Especially in the low tPSA range of 2–4 ng/mL, this advantage for PSAD was clear (Fig. 1A) and confirmed the advantage of PSAD at high sensitivity levels. When comparing PSAD with tPSA, all AUC comparisons (see Fig. 1A–D) reached significance (P < 0.0001). The %fPSA also reached significance compared with tPSA in all tPSA ranges (P < 0.0001), with exception of the 2–4 ng/mL tPSA range (P = 0.17). Looking separately at data from 1996 to 2001, PSAD also was significantly better than %fPSA in the 4–10 ng/mL tPSA range (P < 0.01). However, data from 2001 to 2004 revealed that PSAD reached significance compared with %fPSA only in the tPSA range of 2–4 ng/mL (P < 0.01).
|tPSA range||AUC 1996–2004||AUC 1996–2001||AUC 2001–2004|
|2–20 ng/mL||0.664 ± 0.014a||0.755 ± 0.012ab||0.777 ± 0.011||0.678 ± 0.016a||0.736 ± 0.015ab||0.782 ± 0.014||0.660 ± 0.024a||0.797 ± 0.019b||0.807 ± 0.019|
|2–4 ng/mL||0.644 ± 0.033a||0.667 ± 0.033a||0.739 ± 0.03||0.619 ± 0.045a||0.639 ± 0.041a||0.746 ± 0.036||0.728 ± 0.057c||0.72 ± 0.058c||0.80 ± 0.049|
|2–10 ng/mL||0.643 ± 0.016a||0.732 ± 0.014ab||0.766 ± 0.013||0.658 ± 0.019a||0.709 ± 0.018ab||0.767 ± 0.016||0.634 ± 0.028a||0.778 ± 0.022b||0.80 ± 0.021|
|4–10 ng/mL||0.569 ± 0.019a||0.731 ± 0.017b||0.744 ± 0.016||0.569 ± 0.024a||0.702 ± 0.022bc||0.734 ± 0.021||0.569 ± 0.033a||0.783 ± 0.025b||0.79 ± 0.025|
|10–20 ng/mL||0.528 ± 0.031a||0.779 ± 0.023b||0.77 ± 0.025||0.543 ± 0.037a||0.758 ± 0.028b||0.775 ± 0.029||0.515 ± 0.061a||0.815 ± 0.043b||0.787 ± 0.046|
Finally, time-dependent observations are worth mentioning. For the years 2001–2004, we found a significant increase in prostate volume in patients with BPH in the tPSA ranges of 2–4 ng/mL, 4–10 ng/mL, 2–10 ng/mL, and 2–20 ng/mL (P values from 0.002 to < 0.0001) compared with the patients with BPH from 1996 to 2001. In patients with PCa, the behavior was similar. A significant increase for the latest years (2001–2004) in prostate volume was observed for the tPSA ranges 2–10 ng/mL, 2–20 ng/mL, 4–10 ng/mL, and 10–20 ng/mL (P values from 0.005 to < 0.0001) compared with the patients from 1996 to 2001. The percentage of patients with suspicious DRE results was significantly higher in patients with PCa in the first years (1996–2001) compared with the recent years (2001–2004) for all tPSA ranges analyzed with exception of the 2–4 ng/mL tPSA range. In patients with BPH, we did not find any differences.
Although the determination of tPSA is recognized as the best diagnostic tool for the early detection of PCa, the sensitivity and specificity of tPSA is not sufficient. When PSA alone is used to predict the probability of PCa within the 4–10 ng/mL range, approximately 75% of all biopsies will be negative.20, 37
After the introduction by Benson et al.19, 20 of PSAD as a useful tool to increase specificity, early studies already could demonstrate the advantage of PSAD compared with tPSA.38–40 However, others did not confirm this advantage and reported only equal results for PSAD and tPSA.41, 42
An important step in the further improvement of tPSA specificity was the measurement of fPSA and the use of %fPSA.4, 5 The clinical advantage of %fPSA, compared with tPSA, was demonstrated in the 4–10 ng/mL tPSA range.7, 9, 26, 37, 43 Data from this study on 975 patients with tPSA in the range of 4–10 ng/mL confirmed the well known correlation. Both %fPSA and PSAD were significantly better than tPSA but performed equally regarding the AUC (0.731 vs. 0.744, respectively) and the cut-off levels at 90% and 95% sensitivity (39% vs. 36% and 20% vs. 25% specificity). All studies with ≥ 500 patients and available data on men with tPSA in the 4–10 ng/mL range demonstrated a similar performance of PSAD and %fPSA.16, 17, 26, 30, 31 Thus, if PSAD does not perform better than %fPSA, then it cannot replace %fPSA, although it may be used as additional tool for %fPSA to detect PCa. The opportunity to perform an (uncomfortable) TRUS and calculating the PSAD, therefore, should be used especially in patients with lower tPSA concentrations, in whom the %fPSA cannot help to decide on a biopsy. Compared with an often unnecessary prostate biopsy, the costs and discomfort of TRUS are acceptable for patients who have borderline tPSA and %fPSA values. A new approach has been published recently in which the authors calculated the ratio of %fPSA to prostate volume and called it the %fPSA density.33 For the 105 men (20 men with PCa) in that study, the %fPSA density reached an AUC of 0.771, and PSAD reached an AUC of 0.75 with no difference but with a significant better performance compared with %fPSA (AUC, 0.604).33 A further improvement for better discrimination between PCa and BPH are the so-called artificial neural networks, which usually use both parameters, %fPSA and prostate volume (PSAD), in conjunction with age, tPSA, and other factors, like DRE status.37, 44, 45 Those studies consistently demonstrated a significant improvement compared with %fPSA. In the current study, we did not separate nonsuspicious and suspicious DRE results and analyzed ROC curves separately because of the large individual variability of the DRE results and the differences between investigators. It was not our objective to exclude patients from the overall analysis only because a likely suspicious DRE results was reported. However, in our free, useable artificial neural network, the factor DRE status already is included.37
At lower tPSA concentrations (< 4 ng/mL), there are controversial data regarding the %fPSA compared with tPSA. Some studies demonstrated an advantage for %fPSA,25, 46–48 and some did not demonstrate a better performance by %fPSA compared with tPSA.49–52 Retrospective data obtained by Jung et al.48 revealed an advantage for %fPSA compared with tPSA. In a larger group of men (n = 219 patients) with tPSA in the 2–4 ng/mL range, there was no difference between tPSA and %fPSA (AUC, 0.62 vs. 0.64).37 Data from the current study also showed no advantage for the use of the %fPSA at a low tPSA range of 2–4 ng/mL in the 307 investigated men. However, the number of patients and the proportion of patients with PCa, which was high in the current study (63,4%), may have had an effect on ROC analyses. This has to be considered, especially when analyzing cut-off levels for positive and negative predictive values. Sensitivity and specificity cut-off levels, like those used in the current study, should not be influenced by the proportion of patients with PCa. One of the main results of the current study is that PSAD showed a better performance than %fPSA at tPSA concentrations < 4 ng/mL. Our recommendation, therefore, is to perform a TRUS instead of or after the free PSA measurements if the PSA level is between 2 ng/mL and 4 ng/mL to lower the number of unnecessary biopsies. The AUC was significantly larger for PSAD (AUC, 0.739) compared with %fPSA (AUC, 0.667) when all patients in this tPSA range were analyzed. This significance also was observed by looking separately at the data from 1996–2001 and 2001–2004. To confirm this advantage of PSAD compared with %fPSA at low tPSA concentrations, investigations in large screening populations should be performed, because our population did not represent a screening population. At 95% and 90% sensitivity, PSAD with specificities of 21% and 35% also performed significantly better than %fPSA with 7% and 16% specificities. It also is acceptable to use high specificity cut-off levels for biopsy decisions in patients with low tPSA concentrations (< 4 ng/mL) to avoid many biopsies in general.37, 44 In the current study, PSAD did not reach significance compared with %fPSA but showed higher sensitivities (Table 3). Djavan et al.25 compared %fPSA and PSAD in a large population with tPSA concentrations < 4 ng/mL and found a significantly larger AUC for %fPSA compared with PSAD. The PSAD of the transition zone (PSA-TZ) is another parameter and may show a further improvement in specificity compared with PSAD.25, 26, 53 However, in the current study, we did not estimate PSA-TZ; because, in patients with small glands, in which this zone often is not enlarged markedly, the transition zone is more difficult to measure reliably. In contrast, Gohji et al.54 found in Japanese men that PSA-TZ could not offer additional information that was useful in the detection of PCa, although PSAD does offer such information. Another study also found no advantage of PSA-TZ compared with PSAD.55 Thus, even the measurement of total prostate volume based on TRUS is subject to operator-dependent variability.
The comparison of %fPSA and PSAD at tPSA concentrations of 10–20 ng/mL indicated a slightly better performance of %fPSA compared with PSAD, although there were no significant differences regarding the AUC (0.779 vs. 0.77) and specificities at the sensitivity cut-off levels. This demonstrates the limitation of PSAD with higher tPSA concentrations and its inability to improve the specificity of %fPSA further. However, the use of %fPSA to discriminate between PCa and BPH is limited in patients with large prostates due to the coexistence of both diseases if the patients with PCa have large glands.11, 12 In these patients, PSAD also may be used as an alternative, although the mean values for PCa and BPH narrowed with larger gland volume in the current study. It should be noted that the need for TRUS is a limitation of PSAD compared with %fPSA with regard to time, cost, and discomfort.
The analysis of PSAD mean and 95% sensitivity cut-off levels revealed an important tendency. The higher the tPSA range, the higher the PSAD cut-off levels. Whereas, in the tPSA range 2–4 ng/mL, a cut-off PSAD of 0.05 should be used to obtain 95% sensitivity, this cut-off level must be doubled to 0.1 in the tPSA range 4–10 ng/mL to reach the same sensitivity. For higher tPSA concentrations of 10–20 ng/mL, this cut-off PSAD is 0.19. These results are in contrast to most other studies, which usually preferred PSAD cut-off levels near 0.15.24, 38, 42 In another comparable study regarding the number of patients (n = 1913 patients), Kuriyama et al.29 recommended different reference values for PSAD according to prostate volume. It can seen in Table 2 that our data confirm the lower PSAD values with increasing volume. For cut-off recommendations, the consideration of both tPSA and prostate volume as factors in PSAD probably would work best. This may be the best possible approach with the advent of artificial neural networks, which are capable of weighing the influence of each factor in determining the final result.56
Data from the two different periods showed increasing prostate volume in patients with PCa and BPH within the last years. Because no method was changed, it is difficult to explain this result. However, because more patients are treated with α blockers, this allows the prostate size to grow without urinary symptoms. It is known that PSAD looses its power with increasing prostate volume. This may explain the decreased discriminatory power of PSAD within the second period analyzed (2001–2004).
To summarize this study, the results showed that PSAD increases specificity compared with %fPSA, especially at tPSA concentrations < 4 ng/mL, and that tPSA adapted cut-off values should be used for PSAD. Especially among patients who have low tPSA concentrations, the additional use of PSAD may improve patient selection for prostate biopsy.
- 5A complex between prostate-specific antigen and alpha 1-antichymotrypsin is the major form of prostate-specific antigen in serum of patients with prostatic cancer: assay of the complex improves clinical sensitivity for cancer. Cancer Res. 1991; 51: 222–226., , , , , .
- 17Can indexes based on PSA determine which patients should undergo repeated ultrasound-guided transrectal prostatic biopsy? Study on 546 patients who underwent repeated biopsy]. Arch Esp Urol. 2002; 55: 225–234., , , , , . [
- 34Prostate cancer detection by prostate-specific antigen-related parameters. Hinyokika Kiyo. 2003; 49: 405–410., , , , .
- 41Prospective evaluation of prostate specific antigen (PSA), PSA density, free-to-total PSA ratio and a new formula (prostate malignancy index) for detecting prostate cancer and preventing negative biopsies in patients with normal rectal examinations and intermediate PSA levels. Int Urol Nephrol. 1999; 31: 497–509., , , , , .
- 43Comparison of percent free prostate specific antigen and prostate specific antigen density as methods to enhance prostate specific antigen specificity in early prostate cancer detection in men with normal rectal examination and prostate specific antigen between 4.1 and 10 ng/ml. J Urol. 1997; 158: 502–504., , , et al.
- 47Use of percentage of free prostate-specific antigen to identify men at high risk of prostate cancer when PSA levels are 2.51 to 4.0 ng/mL and digital rectal examination is not suspicious for prostate cancer: an alternative model. Urology. 1999; 54: 220–224., , , et al.
- 48Molecular forms of prostate-specific antigen in serum with concentrations of total prostate-specific antigen < 4 microg/L: are they useful tools for early detection and screening of prostate cancer? Int J Cancer. 2001; 93: 759–765., , , et al.
- 50Percent free prostate specific antigen in the total prostate specific antigen 2 to 4 ng/ml range does not substantially increase the number of biopsies needed to detect clinically significant prostate cancer compared to the 4 to 10 ng./ml. range. J Urol. 2002; 168: 504–508., , .
- 52Ratio free/total PSA (%f-PSA) in men with total PSA 0.5–3.0 ng/ml—Improvement for prostate cancer detection? J Urol. 2000; 163(Suppl): 277A., , , , , .
- 54Detection of prostate carcinoma using prostate specific antigen, its density, and the density of the transition zone in Japanese men with intermediate serum prostate specific antigen concentrations. Cancer. 1997; 79: 1969–1976., , , et al.
- 56An artificial neural network considerably improves the diagnostic power of percent free prostate-specific antigen in prostate cancer diagnosis: results of a 5-year investigation. Int J Cancer. 2002; 99: 466–473., , , et al.