BJU International (1999), 83, 1000–1002

Free prostate-specific antigen ‘in the field’: a useful adjunct tostandard clinical practiceG.N. COLLINS, K. AL EXANDROU, A. WYNN-DAVIES, S. MOBLEY and P.H. O’REILLYDepartment of Urology, Stepping Hill Hospital, Stockport, UK

Objectives To determine the clinical utility of the ratio Results PSA levels, PSA density and fPSA density diCeredsignificantly between those with benign histology andof free to total prostate-specific antigen (f/tPSA), which

may improve the discrimination of PSA testing in the neoplasia. The f/tPSA only diCered significantly inmen with a PSA level of 10.1–15 ng/mL. The f/tPSAassessment of patients with prostatic disease, in a busy

clinical practice. threshold of 0.25 was the most useful clinically, witha negative predictive value for prostate cancer of 91%.Patients and methods A series of 198 men undergoing

transrectal ultrasonography (TRUS)-guided biopsy, Using this threshold would reduce the negative biopsyrate by 30%.because of a high total PSA level or an abnormal

digital rectal examination or both, had blood samples Conclusion The f/tPSA is of clinical value in reducingthe negative biopsy rate and in the management oftaken for the assessment of both free and total

PSA before any form of prostatic manipulation. The patients with a high PSA level and previous negativebiopsies.histological findings were compared with tPSA, fPSA

and f/tPSA, evaluating three diCerent thresholds of Keywords Prostate specific antigen, PSA, free PSA, ratio,prostate cancer, biopsyf/tPSA.

measured using a standardized reproducible techniqueIntroduction

[3]. The PSA characteristics of these patients with histo-logically confirmed benign and malignant prostateOnly 7 years has elapsed since Lilja et al. [1] and

Stenman et al. [2] first described the characteristics of glands were analysed retrospectively. The men weregrouped on the basis of their PSA level into four arbitraryPSA in the circulation, showing that it is largely bound

to alpha-1-antichymotrypsin (85%) with a small categories and the sensitivity and specificity values ofvarious recommended f/tPSA thresholds were deter-unbound inactive portion (15%). These groups also

noted that the median ratio of free unbound PSA (fPSA) mined. Two sample t-tests with 95% CIs were used tocompare the mean PSA and fPSA, and the mean ratioto total (t)PSA (f/tPSA) in men with BPH was higher

than in men with prostate cancer (0.28 vs 0.18), leading of the f/tPSA.to many reports of the possible value of f/tPSA inenhancing the discrimination of PSA testing in the

Resultsdiagnosis and management of patients with potentialprostate cancer. The present report describes our experi- The main indication for referral for TRUS was an elevated

PSA level (> 4 ng/mL) and 57 of the 198 men (29%)ence with f/tPSA in a busy urological unit of a districtgeneral hospital. were subsequently diagnosed with prostate cancer. The

distributions of PSA, fPSA and f/tPSA according tohistology are shown in Table 1. PSA and f/tPSA diCered

Patients and methodssignificantly between those with benign and malignantprostates, although the fPSA diCerence was less signifi-A series of 198 men referred for TRUS and prostatic

biopsy on clinical grounds were assessed with fPSA and cant. The PSA and fPSA densities were significantlygreater in those with cancer than in those with benigntPSA levels before undergoing any form of prostatic

manipulation. PSA levels were measured using the disease (Table 1). When subdivided according to PSAlevel, the arithmetic mean f/tPSA was less in those withappropriate Tandem-R assays (Hybritech Inc., San Diego,

CA, USA) on one site. The fPSA result did not influence cancer but was only significant for those with a PSAlevel of 10.1–15 ng/mL (Table 1). The sensitivity,the decision to perform biopsy. Prostate volume wasspecificity, positive (PPV) and negative predictive value(NPV) in cancer diagnosis using f/tPSA thresholds ofAccepted for publication 9 February 1999

1000 © 1999 BJU International

FRE E PSA ‘IN THE FIELD’ 1001

Table 1 The various estimates of PSA in patients with benign disease or cancer. Values are the mean with 95% CI

HistologyVariable Benign N Cancer N

Median PSA (ng/mL);total 6.96 (6.16–7.86) 131 12.83 (10.05–16.37)‡ 55Free 1.31 (1.17–1.48) 124 1.62 (1.23–2.14) 53f/t 0.22 (0.20–0.24) 124 0.15 (0.13–0.17)‡ 53tPSA density 0.159 (0.139–0.183) 129 0.378 (0.286–0.500)‡ 51fPSA density 0.03 (0.027–0.034) 122 0.048 (0.036–0.065)† 49

Mean f/t PSA for tPSA group∏4 0.32 (0.25–0.39) 30 0.21 (0.16–0.25) 64.1–10 0.20 (0.18–0.22) 64 0.17 (0.13–0.21) 1710.1–15 0.17 (0.15–0.20) 20 0.11 (0.07–0.14)† 10�15 0.16 (0.09–0.22) 10 0.13 (0.10–0.16) 20

†P < 0.01, ‡P < 0.001, Student’s t-test.

0.12, 0.18 and 0.25 are listed in Table 2; the diagnostic Finasteride therapy lowers tPSA and fPSA levels butdoes not alter the f/tPSA [16]. Neither does the f/tPSAaccuracies of an arbitrary PSA density threshold of 0.25

are also listed. change with age [17] and therefore any decrease inf/tPSA over time might be important. Some investigatorsare examining fPSA density and fPSA/transition zone

Discussiondensity, as we and others have shown that fPSA levelsare likely to be largely dependent on the transition zoneThere have been many recommendations for the clinical

application of fPSA and f/tPSA; the use of the ratio may [18,19]. Also, the release of fPSA after manipulationdiCers according to histology [18] and whether this hasreduce the number of unnecessary biopsies by a third or

more, particularly in those men with a PSA of clinical applications is currently under investigation inthis unit (unpublished data).4–10 ng/mL [4–6] and may increase the specificity of

PSA by 24% [7]. It may be of particular value in those We recommend using a threshold f/tPSA of > 0.25in clinical practice; this threshold would have reducedpatients with rising PSA levels and negative biopsies [8],

and indeed most investigators have found it useful. The the present negative biopsy rate by 30%, saving#£20 000 annually and the morbidity of the biopsy forratio may be useful in the community setting, depending

on the threshold value used [9], while others have the patient. Four cancers would not have been detected(6%), which we consider an acceptable ‘miss’ rate, asquestioned its value in predicting tumour stage [10,11].

Two UK studies considered fPSA to be of no clinical we monitor any patients with borderline PSA levelsusing PSA velocity rather than single measurements.value; one included relatively few patients while the

other, in addition to having few patients, based the The f/tPSA is particularly valuable in those patients withpersistently raised PSA levels and previous benign biops-results largely on ROC analysis [12,13]. Other studies

suggest that a measurement of complexed PSA may be ies. We use a combination of PSA velocity with thef/tPSA negative velocity (Fig. 1) in the decision tomore useful than fPSA [14,15]. The main problem with

studies of f/tPSA is the lack of consensus about an re-biopsy these men, e.g. a patient with a persistentlyelevated steady PSA level and a falling f/tPSA wouldappropriate threshold and hence the variation in

reported results. undergo re-biopsy in our unit. The f/tPSA was mar-ginally more useful in diagnosing cancer than was PSAHowever, there are other areas of potential value.

Table 2 The diagnostic accuracy of thevarious estimates of PSA in patients withbenign disease or cancer

Statistic Threshold f/tPSA PSA densityn/N (%) ∏0.12 ∏0.18 ∏0.25 ∏0.25

Sensitivity 23/53 (43) 37/53 (70) 49/53 (92) (69)Specificity 105/124 (85) 69/124 (56) 39/124 (31) (78)PPV 23/42 (55) 37/92 (40) 49/134 (37) (55)NPV 105/135 (78) 69/85 (81) 39/43 (91) (86)

© 1999 BJU International 83, 1000–1002

1002 G.N. COLLINS et al.

5 Partin AW, Catalona WJ, Southwick PC, Subong EN, GasiorGH, Chan DW. Analysis of percent free prostate-specificantigen (PSA) for prostate cancer detection: influence oftotal PSA, prostate volume and age. Urology 1996;48: 55–61

6 Smith DS, Catalona WJ, Keetch DW. Comparison of percentfree PSA and PSA density as methods to enhance thespecificity of PSA screening. J Urol 1996; 155: 422A

7 Oesterling JE, Wonjo KJ, England B et al. A comparison offree to total PSA (F/T) ratio to total PSA for distinguishingbenign prostatic hyperplasia (BPH) from prostate cancer(CaP) using the Abbott AxSYM system. J Urol 1996;155: 370A

8 Morgan TO, McLeod DG, Moul J, Connelly RR, DouglasTH, Murphy GP. Clinical use of free PSA to avoid repeatTime

f/t

PS

A o

r P

SA

prostate biopsies in men with elevated total PSA. J UrolFig. 1. PSA velocity and fPSA negative velocity, showing four1996; 155: 370Atypes of response; both stable (light green and light red), suggesting

9 Klee GG, Lerner SE, Jacobsen SJ et al. Predictive power ofno biopsy; a stable ratio (light red) and rising PSA level (darkfree: total PSA ratio is not superior to total PSA in thegreen), i.e. consider no biopsy but an early review; a stable PSA

level (dark green) and falling ratio (dark red), leading to biopsy; a diagnosis of prostate cancer in the community setting.rising PSA level (light green) and falling ratio (dark red), leading J Urol 1996; 155: 371Ato biopsy. 10 Partin AW, Subong ENP, Jones KA et al. Free/total PSA

does not improve the prediction of final pathological stagedensity in the present study. The measurement of fPSA for men with localized prostate cancer. J Urol 1996;is more economical than TRUS, although a combined 155: 415Aanalysis of the diagnostic accuracy of f/tPSA and density 11 Graefen M, Hammerer P, Henke P, Hilz H, Huland E,

Huland H. Percent of free PSA does not correlate withcombined would be interesting.pathological outcome. J Urol 1996; 155: 369AIn conclusion, the present results show diCerent PSA

12 Masters JG, Keegan PE, Hildreth AJ, Greene DR. Free/totalcharacteristics in men with benign and malignant pros-serum prostate-specific antigen ratio: how helpful is it intates, and confirm the clinical value of measuring f/tPSA.detecting prostate cancer? Br J Urol 1998; 81: 419–23We recommend that it is used: (i) to decrease the overall

13 Crundwell MC, Cooke PW, Cramb R et al. Does the use ofnegative biopsy rate (with significant economic andpercentage free PSA decrease the negative biopsy rate

morbidity benefits); (ii) to aid the decision to re-biopsyin men with suspected prostate cancer? Br J Urol 1998;

patients with mild and moderately increased tPSA levels 81: 67who have previous benign histology; and (iii) to monitor 14 Brawer MK, Meyer GE, Letran JL et al. Measurement ofmore complex patients, e.g. those with multiple previous complexed PSA improves specificity for early detection ofbenign biopsies or comorbidity, using the combination prostate cancer. Urology 1998; 52: 372–8

15 Sokoll LJ, Bruzek DJ, Cox JL, Partin AW, Chan DW. Isof PSA velocity and f/tPSA negative velocity.complexed PSA alone clinically useful? J Urol 1998; 159

(Suppl): 234References 16 Pannek J, Marks LS, Pearson JD et al. Influence of finasteride

1 Lilja H, Christennson A, Dahlen U et al. Prostate specific on free and total serum prostate specific antigen levels inantigen in serum occurs predominantly in complex with men with benign prostatic hyperplasia. J Urol 1998;alpha1-antichymotrypsin. Clin Chem 1991; 37: 1618 159: 449–53

2 Stenman UH, Leinonen J, Alfthan H, Rannikko S, 17 Oesterling JE, Jacobsen SJ, Klee GG et al. Free, complexedTukhanen K, Alfthen O. A complex between prostate and total serum prostate specific antigen: the establishmentspecific antigen and alpha 1-antichymotrypsin is the major of appropriate reference ranges for their concentrationsform of prostate specific antigen in serum of patients with and ratios. J Urol 1995; 154: 1090–5prostate cancer, assay of the complex improves clinical 18 Collins GN, Martin PJ, Wynn-Davies A, Brooman P, O’Reillysensitivity for cancer. Cancer Res 1991; 51: 222–6 PH. The eCect of digital rectal examination, flexible

3 Collins GN, Raab GM, Hehir M, King B, Garraway WM. cystoscopy and prostatic biopsy on free and total prostate-Reproducibility and observer variability of transrectal specific antigen, and the free-to-total prostate specificultrasound measurements of prostatic volume. Ultrasound antigen ratio in clinical practice. J Urol 1997; 157:Med Biol 1995; 21: 1101–5 1744–7

4 Catalona WJ, Smith DS, Wolfert RL et al. Evaluation of 19 Stamey TA, Yang N, Hay AR, McNeal JE, Freiha FS,percentage of free serum prostate specific antigen to Redwine E. Prostate-specific antigen as a serum marker forimprove specificity of prostate cancer screening. JAMA adenocarcinoma of the prostate. New Engl J Med 1987;

317: 9091995; 274: 1214

© 1999 BJU International 83, 1000–1002