EP1068534A1

EP1068534A1 - Method for differentiation of prostate cancer

Info

Publication number: EP1068534A1
Application number: EP99902563A
Authority: EP
Inventors: Maciej Kwiatkowski; Jonas Hugosson; Hans Lilja; Timo Lövgren; Kim Pettersson; Timo Piironen; Franz Recker
Original assignee: Arctic Partners Oy AB
Current assignee: Arctic Partners Oy AB
Priority date: 1998-03-04
Filing date: 1999-02-03
Publication date: 2001-01-17
Also published as: WO1999045398A1; JP2002506210A; FI980488A; FI980488A0

Abstract

The invention concerns a method for differentiating patients with cancer of the prostate (PCa) from patients with benign prostatic hyperplasia (BPH) and healthy male subjects without PCa, wherein the individual's body fluid concentration of human glandular kallikrein 2 (hK2) and prostate specific antigen (PSA) have been determined. According to the invention, a combination of hK2 and PSA, wherein PSA means the free PSA, the complexed PSA or the total PSA, is used as marker distinguishing PCa patients from BPH and other non-PCa individuals. Furthermore, the invention relates to a method for differentiating patients with indolent or slowly progressive prostate cancers from patients with aggressively progressive prostate cancers.

Description

METHOD FOR DIFFERENTIATION OF PROSTATE CANCER

FIELD OF INVENTION

This invention relates to the differential diagnosis of prostate cancer from healthy controls or benign prostatic conditions by combination of results obtained by an analytically sensitive and specific immunoassay for human glandular kallikrein 2 (hK2) with those obtained by immunoassays specifically detecting various forms of prostate specific antigen (PSA or hK3 ) , most notably the free form of PSA. The invention concerns the diagnosis of prostate cancer both in screening of asymptomatic individuals as well as in distinguishing between cancer and benign conditions in men presenting with clinical symptoms. The invention reveals diagnostic improvements in that higher clinical sensitivities and/or specificities can be obtained both in relation to the conventionally used determination of total PSA, as well as to the more recently introduced method of determinating the proportion of complexed or free PSA to total PSA. Suitable biological specimens for the immunoassay determinations are serum, plasma or whole blood samples.

INTRODUCTION AND BACKGROUND

The publications and other materials used herein to illuminate the background of the invention, and in particular, cases to provide additional details respecting the practice, are incorporated by reference.

Human glandular kallikrein 2 (hK2) and prostate-specific antigen (PSA; also designated hK3 ) are two closely related serine proteases highly expressed predominantly in prostatic tissue. [Wang et al . Invest Urol 1979; 17:159- 163, Chapdelaine et al . FEBS Lett 1988; 236:205-208]. They show extensive similarities in their amino-acid sequence (79%) but the expression rates are quite different (hK2 mRNA levels amount to ~ 10-20% of PSA rtiRNA levels) [Schedlich et al . DNA 1987; 6:429-437]. Enzymatically active PSA is secreted into seminal fluid at high concentrations (0.2-5 mg/mL) [Christensson et al . Eur J Biochem 1990; 194:755-763, Ahlgren et al . 1995 J Androl 16:491-498]. In semen, PSA degrades the seminal vesicle derived gel-forming proteins causing liquefaction of semen and release of progressively motile spermatozoa [Lilja J Clin Invest 1985; 76:1899-1903]. Recently, recombinant hK2 was shown to convert in vi tro inactive recombinant proPSA into active mature PSA [Lovgren et al . Biochem Biophys Res Commun 1997; 238:549-555, Takayama et al. J Biol Chem 1997; 272:21582-21588, Kumar et al . Cancer Res 1997; 57:3111- 3114].

The incidence of prostate cancer has increased during the last decade mainly due to prolonged lifetime and increased screening. This fact underlines the need of improved diagnostic approaches and new treatments. Analysis of PSA in serum is well established in the diagnosis and monitoring of prostate-cancer (PCa) patients [Oesterling J Urol 1991; 145:907-923]. However, raised serum concentrations of PSA are also found in patients with other prostatic diseases, for instance benign hyperplasia of the prostate (BPH) [Hudson et al . J Urol 1989; 142:1011-1017]. PSA has been shown to possess chymotrypsin-like protease activity [Lilja et al . J Biol Chem 1989; 264:1894-1900 Christensson et al . Eur J Biochem 1990; 194:755-763). The active single-chain form of PSA forms stable covalent complexes with several extracellular protease inhibitors, such as αi-antichymotrypsin (ACT), ₂-macroglobulin (AMG), pregnancy-zone protein (PZP), protein C inhibitor (PCI), and cti-antitrypsin [Christensson et al . Eur J Biochem 1990; 194:755-763, Stenman et al . Cancer Res 1991; 51:222-226, Espana et al . Thromb Res 1991; 64:309-320, Christensson and Lilja Eur J Biochem 1994; 220:45-53, Zhang et al . Prostate 1997; 33:87-96]. The discovery of several different molecular forms of PSA in serum have improved the specificity for PCa to a certain extent. In blood, the predominant form of immunodetectable PSA is in complex with ACT and only a minor fraction is in a free form (PSA-F) [Stenman et al . Cancer Res 1991; 51:222-226, Lilja et a l . Clin Chem 1991; 37:1618-1625]. In general, patients with BPH have a higher proportion of PSA-F in serum than patients with PCa. This has led to the use of the ratio of PSA-F to PSA-T (PSA-T = PSA-F + PSA-ACT + other quantitatively less important PSA-serpin complexes) to distinguish between BPH and PCa in case of moderately raised PSA levels [Stenman et al . Cancer Res 1991; 51:222- 226, Christensson et al . J Urol 1993; 150:100-105]. Although this has improved the specificity for PCa, there is still a great need for markers which provide further improved discrimination of cancer patients from normals and benign conditions .

hK2 is also expressed at fairly high levels in normal prostate epithelia [Chapdelaine et al . FEBS Lett 1988; 236:205-208]. In adenocarcinoma of the prostate, hK2 is expressed at even higher levels [Darson et al . Urology 1997; 49:857-862.]. hK2 can be detected in seminal plasma [Deperthes et al . Biochim Biophys Acta 1995; 1245:311- 316.], but it has not been purified (in large quantities) or characterized thoroughly. The physiological function of hK2 is not yet known, but the primary structure of hK2 suggests it is likely to possess a trypsin-like catalytic activity [Schedlich et al . DNA 1987; 6:429-437].

Recombinant production has made it possible to further study the characteristics of hK2 [Eerola et al . Prostate 1997; 31:84-90]. The close similarity in amino acid sequence to PSA makes cross-reactivity likely in PSA immunoassays based on anti-PSA antibodies . We have previously studied the cross reactivity of 23 different anti-PSA Mabs with recombinant hK2 [Lδvgren et al . Biochem Biophys Res Commun 1995; 213:888-895], only five of them were found to cross-react with hK2 and with similar affinities as those for PSA. These findings have recently been confirmed by us in a further extended and detailed epitope mapping study [Piironen et al . Protein Science 1998; 7:259-269]. An immunoassay for hK2, using a blocking procedure with anti-PSA Mabs that do not crossreact with hK2 followed by detection with anti-PSA Mabs that fully crossreact with hK2 have been constructed and found to be specific and sensitive for hK2 in patient serum samples (detection limit 0.1 μg/L and crossreactivity with PSA 0.7%) [Piironen et al . Clin Chem 1996; 42:1034-1041]. Applied to a panel of clinically non-defined patient samples it was concluded that 1) serum hK2 and total PSA are correlated to each other over the whole PSA range (r=0.84 for PSA range 1 to 3400 μg/L), but in a more restricted range this correlation was poor (r=0.32 for PSA range 1 to 20 μg/L). 2) The concentrations of hK2 were generally low in comparison to PSA (only in 7 % of the cases did hK2 exceed 5 % of total PSA) . These results emphasize the critical importance of two analytical criteria i.e. the crossreactivity with PSA and the lower limit of detection, of any hK2 assay to be applied to clinical specimens . Gel filtration of some serum samples with high hK2 concentrations showed that the immunodetectable hK2 eluted at a size of approximately 30 kDa indicating that immunodetectable hK2 is in a free form. In this study, we have further validated and improved the critical performance characteristics of this hK2 assay and applied it to sera from several carefully characterized patient groups in order to study its clinical utility in urological practice. Comparisons to assays simultaneously performed for PSA-T, PSA-F and PSA-ACT are reported in three different groups of well defined clinical specimens.

SUMMARY OF THE INVENTION AND PREFERRED EMBODIMENTS

This invention relates to the use of sensitive immunoassays specific for human glandular kallikrein- hK2 for the differential diagnosis of prostate cancer from healthy control subjects or from subjects suffering from benign conditions of the prostate.

The immunoassays of hK2 can be performed on body fluids such as serum, plasma or whole blood samples obtained from the individuals under investigation.

According to the central findings of this invention, the use of hK2 as an efficient tumor marker is best realized by combining the measured hK2 concentrations with measured concentrations of different forms of prostate specific antigen (total PSA, compexed PSA, free PSA), most notably the free form of PSA.

Thus, according to one aspect, this invention concerns a method for differentiating patients with cancer of the prostate (PCa) from patients with benign prostatic hyperplasia (BPH) and healthy male subjects without PCa, wherein the individual's body fluid concentration of human glandular kallikrein 2 (hK2) and prostate specific antigen (PSA) have been determined. The method is characterized in that the combination of hK2 and PSA, wherein PSA means the free PSA, the complexed PSA or the total PSA, is used as marker distinguishing PCa patients from BPH and other non- PCa individuals .

According to one preferred embodiment, the marker is the ratio hK2/F or the ratio F/hK2, wherein F is the concentration of free PSA.

According to another preferred embodiment, the marker is the ratio hK2/F or the ratio F/hK2 multiplied by a quantity characterizing the PSA concentration. The term "quantity characterizing the PSA concentration" can be free PSA, total PSA or complexed PSA (i.e. PSA complexed with ACT) or any quantity built up from two or all three of said terms (free PSA, total PSA or complexed PSA), such as the ratio "total PSA/free PSA" etc. Thus, according to a particularly preferred embodiment the marker is hK2/F times T ( (hK2 x T)/F), wherein T is the concentration of total PSA. According to another particularly preferred embodiment, the marker is hK2/F times T/F ((hK2/F) x T/F), wherein T is the concentration of total PSA. According to a particularly preferred embodiment, the cut off limit for the ratio is chosen in different ways depending on whether high detection sensitivity or high detection specificity is preferred.

According to a third preferred embodiment, the combination of hK2 and PSA, wherein PSA means the free PSA, the complexed PSA or the total PSA, is performed by logistic regression analysis. According to a particularly preferred embodiment, the logistic regression analysis combination is c[hK2, F], wherein F is free PSA. According to another particularly preferred embodiment, the logistic regression analysis combination is c[hK2, F, T], wherein F is free PSA and T is total PSA.

According to a preferred embodiment of the invention, it is most efficiently applied to a subset of patient samples restricted by intermediate concentrations of total PSA, i.e. concentrations of PSA where the diagnostic discrimination of cancer and non cancer conditions by total PSA alone is more unsecure and in which the cancers - if found - are likely to be organ confined and eligible for curative treatments. Such an area is frequently defined as the total PSA range from 1 to 20 μg/L, preferably 3 or 4 to 10 μg/L.

According to another aspect of the invention, the hK2 containing ratios or combinations through logistic regression analysis analysis of hK2 and other measured parameters can also be used to identify subgroups of the identified prostate cancer patients, more specifically those prostate cancers that are likely to remain indolent or progress slowly (organ confined or localized cancers of low Gleason score or grade) from those cancers that are likely to progress more aggressively (organ confined or localized cancers of high Gleason score or grade).

According to this aspect, the invention thus concerns a method for differentiating patients with indolent or slowly progressive prostate cancers from patients with aggressively progressive prostate cancers, wherein the individual's body fluid concentration of human glandular kallikrein 2 (hK2) and prostate specific antigen (PSA) have been determined. The method is characterized in that the combination of hK2 and PSA, wherein PSA means the free PSA, the complexed PSA or the total PSA, is used as marker distinguishing PCa patients with indolent or slowly progressive prostate cancers from patients with aggressively progressive prostate cancers. The wording "combination of hK2 and PSA, wherein PSA means the free PSA, the complexed PSA or the total PSA" is the same as that discussed above. The marker is preferably the ratio hK2/F or the ratio F/hK2, wherein F is the concentration of free PSA. Alternatively, the marker is preferably the ratio hK2/F or the ratio F/hK2 multiplied by a quantity characterizing the PSA concentration, wherein the "quantity characterizing the PSA concentration" has the same meaning as presented above. Most preferably, the marker is hK2/F times T/F ((hK2/F) x T/F), wherein T is the concentration of total PSA. According to a particularly preferred embodiment, the cut off limit for the ratio is chosen in different ways depending on whether high detection sensitivity or high detection specificity is preferred.

Alternatively, the marker is a combination of hK2 and PSA, wherein PSA means the free PSA, the complexed PSA or the total PSA, said marker being a logistic regression analysis combination. As examples can be mentioned the logistic regression analysis combination is c[hK2, F], wherein F is free PSA, and the logistic regression analysis combination is c[hK2, F, T], wherein F is free PSA and T is total PSA.

BRIEF DESCRIPTION OF THE DRAWINGS

In Figure 1A-1C ROC curves of PSA-T, PSA-F, hK2, PSA-F/PSA- T, hK2/PSA-F, (hK2/PSA-F)x(PSA-T/PSA-F) , c [ PSA-T, PSA-F ] , c [hK2, PSA-F ] , and c[hK2 , PSA-F, PSA-T] are presented for study population 2 (SP2).

In Figure 2A-2C ROC curves of PSA-T, PSA-F, hK2 , PSA-F/PSA- T, hK2/PSA-F, (hK2/PSA-F)x(PSA-T/PSA-F) , c [PSA-T, PSA-F ] , c [hK2, PSA-F ] , and c[hK2 , PSA-F, PSA-T] are presented for study population 3 (SP3).

DETAILED DESCRIPTION OF THE INVENTION

This invention relates to the use of sensitive immunoassays specific for human glandular kallikrein 2 (hK2) for the differential diagnosis of prostate cancer from healthy control subjects or from subjects suffering from benign conditions of the prostate.

According to the invention, the results from an immunoassay specific for hK2 applied to biological samples from patients to be screened ot tested for the presence of malignant or benign prostatic disorders are combined with those obtained from assays of the free fraction of PSA, in order to form a ratio of hK2 to free PSA (or vice versa free PSA to hK2) and identifying a cut-off limit or decision point that provides the best or otherwise desired separation of individuals likely to have prostate cancer, benign prostatic hyperplasia, or are likely to present with no prostatic lesions. The cut-off selected varies depending on whether a high sensitivity or a high specificity is preferred in detecting the prostatic disorder (PCa or BPH).

The mentioned ratio can be supplemented with the concentrations provided by assays specific for total PSA (=free and complexed PSA), assays that measures the sum of total PSA and hK2, assays for PSA (or PSA+hK2 ) complexed to ACT and/or other serpins, or with the ratio of PSA-F to PSA-T (PSA-T can be replaced also by PSA-T+hK2 ) . Such ratios are e.g. hK2xPSA-T/PSA-F or (hK2/PSA-F ) x(PSA-T/PSA- F) - many other modified ratios with improved capability to separate prostate cancer from benign prostatic conditions and healthy conditions can be designed.

The combination of hK2 with PSA-F (and/or with measurement of other forms of PSA or PSA+hK2 ) can be accomplished in another way (not involving formation of ratios or products), but through combination by logistic regression analysis. Combinations by logistic regression analysis such as (hK2, PSA-F) or (hK2, PSA-F, PSA-T) frequently provide even better discrimination than ratios calculated from the individual measurements from each patient as illustrated in the ROC analyses described in the experimental section. Logistic regression analysis is instrumental in providing the basis for various "risk analysis systems that can provide medical decision support" . Other examples of such data handling systems are also: artificial neural networks (ANN), neuro fuzzy networks (NFN) , multilayer perceptron (MLP), learning vector quantization (LVQ) [Freeman et al . In "Neural Networks: Algorithms, Applications and Programming Techniques by Addison-Wesley Publishing

Company" 1991, Zadeh Information and Control 1965; 8:338- 353, Zadeh "IEEE Trans, on Systems, Man and Cybernetics" 1973; 3:28-44, Gersho et al . In "Vector Quantization and Signal Compression by Kluywer Academic Publishers, Boston, Dordrecht, London" 1992, Hassoun "Fundamentals of

Artificial Neural Networks by The MIT Press, Cambridge, Massachusetts, London" 1995].

According to a further aspect of the invention, it is most efficiently applied to a subset of patient samples restricted by intermediate concentrations of total PSA, i.e. concentrations of PSA where the diagnostic discrimination of cancer and non cancer conditions by total PSA alone is more unsecure and in which the cancers - if found - are likely to be organ confined and eligible for curative treatments. Such an area is frequently defined as the total PSA range from 3 or 4 to 10 μg/L, but can be defined differently as regards both the lower limit (which can be even lower) or the higher limit (which can be even higher) .

According to still another aspect of the invention, the hK2 containing ratios or combinations through logistic regression analysis of hK2 and other measured parameters, as shown above for the differential diagnosis of PCa, BPH and other non cancer individuals, can also be used to identify subgroups among the identified prostate cancer patients, more specifically those prostate cancers that are likely to remain indolent or progress slowly from those cancers that are likely to progress more aggressively. The Gleason score is presently the best established predictor (prognostic marker) of a tumors^' pathological potential. Poorly differentiated tumors (high Gleason scores) are characterized by aggressive disease , well differentiated tumors (low Gleason scores) have indolent courses (Cookson et al . J Urol 1997; 157:559).

EXPERIMENTAL

1. Study populations and specimen handling

Study population 1 (SP1). SP1 consisted of 393 consecutive male patients with lower urinary tract symptoms, admitted to the Clinic of Urology, Kantonsspital Aarau, Switzerland, between February 1995 and September 1997. Mean total IPSS symptom score were 20.4 ± 7.3 (obstructive 11.8 ± 4.9 and irritative 8.6 ± 3.2). The inclusion criteria were: no previous prostatic manipulation within 3 weeks, no signs of infection, no chronic catheter prior to admission. These patients were divided into two groups: group I - patients with prostate cancer (PCa), n=104; group II - patients with benign prostatic hyperplasia (BPH), n=289. Median ages (with lower and upper quartiles) were 72 (64, 77) years for the PCa group and 72 (66, 78) years for the BPH group. All patients underwent digital rectal examination (DRE). As a clinical routine, all patients, until the age of 70 years or with a minimal life expectancy of 10 years and a total PSA above 4 ng/mL underwent transrectal ultrasound (TRUS) guided sextant biopsy of the prostate . The same procedure was also applied to patients with suspicious DRE. PCa was revealed by biopsy in 76 cases. From these patients, 29 underwent radical prostatectomy (16 pat. had ρT2, 12 pat. pT3, 1 pat. had pT ) , 34 transurethral resection of the prostate (TURP) (12 cT2, 15 cT3, 7 cT4 ) , 8 received antiandrogen therapy (7 underwent castration by orchiectomy [5 cT4, 2 cT3] and 1 received pharmacological androgen ablation therapy [cT4]). Five were treated conservatively (indwelling Foley catheter or suprapubic catheter) because of contraindications to operation or rescheduled because of signs of infection on admission. In 28 cases PCa was revealed on the histological examination of surgical specimen after TUR-P (12 - pTla, 16 - pTlb). 25 out of 104 patients had metastasized PCa. Of the BPH patients, 270 underwent TUR-P, 3 had transvesical prostatectomy, 16 were treated conservatively (indwelling Foley catheter or suprapubic catheter) because of contraindications to operation or re-scheduled because of signs of infection on admission.

Blood samples were obtained before any prostatic manipulation. After clot formation the samples were centrifuged, serum was collected and frozen at -70 °C. The samples were thawed immediately prior to measurement.

Study population 2 (SP2). SP2 consisted of 115 consecutive patients with lower urinary tract symptoms (mean total IPSS symptom score 18.9 ± 8.0; obstructive 10.7 + 5.3 and irritative 8.2 + 3.2) admitted to the Clinic of Urology, Kantonsspital Aarau, Switzerland, between February 1995 and September 1997. The inclusion criteria were: PSA-T between 3-10 μg/L on admission, no history of previous treatment of prostatic disease, no previous prostatic manipulation within 3 weeks, no signs of infection, no chronic catheter prior to admission. These patients were divided into 2 groups: group I - patients with PCa, n = 25 ; group II - patients with BPH, n=90. Median ages (with lower and upper quartiles) were 70 (62, 72) years for the PCa group and 71 (64, 76) years for the BPH group. All patients underwent DRE. As a clinical routine, all patients until the age of 70 years or with a minimal life expectancy of 10 years underwent TRUS guided sextant prostate biopsies. The same procedure was also applied to patients with suspicious DRE. PCa was revealed by biopsy in 15 cases. From these patients, 6 underwent radical prostatectomy (5 pat. had pT3, 1 pat. pT2c), 6 patients underwent TUR-P (1 cT2, 2 cT3, 1 cT4 ) and 3 patients (3 cT2) were treated conservatively. In 10 cases PCa was revealed upon histological examination of surgical specimens after TUR-P (6 pTla, 4 pTlb). All BPH patients underwent TUR-P.

Blood samples were obtained before any prostatic manipulation. After clot formation, the samples were centrifuged, serum was collected and frozen at -70 °C. The samples were thawed immediately prior to measurement.

Study population 3 (SP3). SP3 consisted of 360 men, aged 51-66 years, participating in a population based prostate cancer screening study in the area of Gothenburg, Sweden, presented initially with a concentration of total PSA > 3 μg/L (Hugosson et al . Submitted). Prior to performing DRE, TRUS and a TRUS guided sextant biopsy, an additional serum sample was obtained. After clotting and centrifugation perfomed within 3 hours, the samples were frozen at -70 °C and thawed immediately before performing the immunoassays . The sextant biopsies revealed prostate cancer in 80 men out of the 360 tested. 44 of these were of clinical grade Tic, 13 T2a, 12 T2b, 4 T2c, 4 T3a, 1 T3b and 2 T4. Gleason scores were determined from histopathological examination of the needle biopsies .

2. Purified proteins, Mabs, reagents, and instrumentation

Recombinant hK2 (rec hK2 ) , recombinant PSA (rec PSA) and purified seminal PSA were obtained using procedures described earlier [Eerola et al . Prostate 1997; 31:84-90, Christensson et al . Eur J Biochem 1990; 194:755-763]. Anti- PSA Mabs H117 and H50 (fully cross-reacting with rec hK2 ) were from Abbot (USA), anti-PSA Mab 2H11 (not cross- reacting with rec hK2 ) was from Wallac Oy (Turku, Finland), and anti-PSA Mabs 36 and 10 (not cross-reacting with rec hK2) were kindly supplied by Olle Nilsson, CanAg (Gothenburg, Sweden). Full description of the binding specificities of the Mabs have been described earlier [Piironen et al . Protein Science 1998; 7:259-269]. Microtitration wells coated with streptavidin, DELFIA¹" 1234 Plate Fluorometer, DELFIA PSA Assay Buffer, DELFIA Wash Solution, DELFIA Enhancement Solution as well as the DELFIA Prostatus™ assay for PSA-T and PSA-F were from Wallac Oy (Turku, Finland) .

3. Immunofluorometric assay procedures for hK2 and forms of PSA

The hK2 assay was based on the previously published assay for hK2 in serum [Piironen et al . Clin Chem 1996; 42:1034- 1041] using the anti-PSA Mabs H50 and H117, with identical affinities for PSA and hK2 and the blocker Mab 2H11 fully specific for PSA [Lδvgren et al . Biochem Biophys Res Commun 1995; 213:888-895, Piironen et al . Protein Science 1998; 7:259-269], but with certain important modifications. In the new assay version, the procedure to block PSA was further inforced by inclusion of two additional PSA- specific anti-PSA Mabs (Mabs 36 and 10). The incubation times were also modified. The new assay protocol was as follows: 25 μL or 50 μL sample or standard were applied in streptavidin coated wells together with 50 μL DELFIA PSA Assay Buffer containing Mabs 2H11, 36 and 10 (2000 ng each/well) and incubated for 1 hour at room temperature. 50 μL PSA assay buffer containing biotinylated Mab H50 (200 ng/well) was added to the reaction mixture and the wells were incubated for 2 hours at room temperature. After a wash step, 200 μL PSA assay buffer containing Eu-labeled tracer Mab H117 (100 ng/well) was added and the wells were incubated for 1 hour at room temperature. Finally, after a second wash step, 200 μL enhancement solution per well was added and the fluorescence measured for 1 s in a plate fluorometer .

The capability of the hK2 assay to equally well (equimolarly) react with complexed as well as free hK2 were studied using in vitro formed complexes of recombinant hK2 with 100 fold molar excess of ACT and protein-C-inhibitor (PCI).

Calibration of the hK2 assay was performed using the PSA calibrators (range 0.2-500 μg/L) from the DELFIA Prostatus EQM commercial kit, but without the use of blocking Mabs. This kit uses the same sandwiching monoclonal antibodies as the hK2 assay. Control samples were rec PSA and serum based in-house controls.

Cross-reaction with PSA as determined from recombinantly produced PSA was less than 0.1%. The analytical detection limit was 0.01 μg/L of hK2. Functional sensitivity (i.e. the concentration at which assay CV:s of replicate determinations were below 20 percent ) was 0.05 μg/L for a sample volume of 25 μL (SP1) and 0.03 μg/L for a sample volume of 50 μL (SP2 and SP3).

The DELFIA Prostatus PSA F/T Dual assay (Wallac, Turku, Finland) using Mab combinations H117/H50 and H117/5A10 was used for detection of PSA-T and PSA-F respectively. The Mab combination H117/H50 measures free PSA and PSA in complex with alpha-1-antichymotrypsin and other serpins in an equimolar fashion [Mitrunen et al . Clin Chem 1995; 41:1115- 1120] and also fully codetects hK2 [Lovgren et al . Biochem Biophys Res Commun 1995; 213:888-895]. Since the proportion of hK2 to PSA in serum is generally very low [Piironen et al . Clin Chem 1996; 42:1034-1041], the concentrations measured by the H117/H50 combination (equal to the T-part of the DELFIA Prostatus PSA F/T Dual assay) are hereafter routinely referred to as PSA-T concentrations. The assay for PSA-F (H117/5A10) does not codetect hK2.

Specific measurement of PSA (and hK2) in complex with ACT was carried out by an investigational two-site immunofluorometric assay based on Mab/Mab combination of H117/241 [Oesterling et al . J Urol 1995; 154:1090-1095].

4. Statistical analyses

Regression analysis was performed for the comparison of PSA-T and PSA-F and PSA-ACT with hK2. Descriptive statistics were given as medians and upper and lower quartiles (25 and 75%-iles). The Kruskal Wallis and Mann- Whitney non-parametric tests were performed to statistically test the differences in PSA-T, PSA-F, hK2, PSA-ACT, PSA-F/PSA-T, hK2/PSA-T, hK2/PSA-F, PSA-F/PSA-ACT and hK2xPSA-T/PSA-F and (hK2/PSA-F) x(PSA-T/PSA-F) between the different groups of patients. Receiver operating characteristics (ROC) plots were used for comparison of the specificity at different sensitivity levels for PSA-T, PSA- F, hK2, PSA-F/PSA-T, hK2/PSA-F and (hK2/PSA-F) (PSA-T/PSA- F) as well as with combinations through logistic regression analysis [Bangma et al . Br J Urol 1997;79:756-762] of PSA-F + PSA-T, PSA-F + hK2 and PSA-F+PSA-T + hK2 , and the areas under the curves (AUC) were calculated with the Windows based GraphROC program, version 2.0 [Kairisto and Poola Scand J Clin Lab Invest Suppl 1995; 222:43-60]. RESULTS

Eσuimolar response of free and complexed hK2 in the hK2 assay

The equimolarity studies of the hK2 assay revealed that hK2 complexed to ACT or PCI relative to free hK2 was almost equally well detected, with 6% negative bias for hK2-ACT, whereas complexation of hK2 to PCI did not produce any discernible effect on the assay response given by the same amount of free hK2 alone.

Results from SP1

The median values and the lower and upper quartiles are compiled in Table 1. All the single parameters measured discriminated between the BPH and PCa with a high statistical significance. The median values of PCa versus those of the BPH group were all increased: 156 percent for PSA-T, 75 percent for PSA-F and 192 percent for hK2. Six ratios formed were also investigated as to their discriminating power, of which PSA-F/PSA-T, PSA-F/PSA-ACT, hK2/PSA-F and the combination of these two ratios into (hK2/F)x(T/F) as well as hK2xPSA-T/PSA-F separated the two groups with high statistical significance. The median percent free PSA was 82 percent higher in BPHs compared to PCa. The median proportion of hK2 relative to PSA-F was 92 percent higher in PCa than in BPH, whereas the median value of the combination of these two ratios were 234 percent higher in PCa than in BPH. In contrast, the median proportion of hK2 relative to PSA-T was not significantly different for BPH and PCa.

In SP1 (patients without any total PSA restriction) hK2 correlated with high significance (p < 0.0001) with PSA-T, PSA-F and PSA-ACT. For the BPH group, the corresponding correlation coefficients (r) were 0.63, 0.77 and 0.57, respectively and 0.73, 0.79 and 0.66 for the PCa group. Results from SP2

Table 2 gives the median values and the lower and upper quartiles of the PCa and BPH groups for the measured variables PSA-T, PSA-F, PSA-ACT and hK2 as well as for ratios PSA-F/PSA-T, hK2/PSA-T, hK2/PSA-F, PSA-F/PSA-ACT, hK2xPSA-T/PSA-F and (hK2/PSA-F) x(PSA-T/PSA-F ) . In forming the ratios, the subjects with hK2 concentrations below the analytical detection limit (PCa n= 1 and BPH n= 1) were given a concentration of equal to the detection limit (0.01 μg/L) . PSA-T values were closely similar for the two groups, 5.7 and 5.6 μg/L for the PCa and BPH groups respectively. Statistical comparison (Mann-Whitney) of the two groups revealed significant discrimination only by two single parameters, PSA-F and hK2. Of the ratios tested, all but hK2/PSA-T resulted in highly significant discrimination of the two groups .

The results from the receiver operating characteristics (ROC) analyses performed for the single parameters and the various ratios formed as well as through combination of (PSA-F, PSA-T), (hK2, PSA-F), and (hK2, PSA-F, PSA-T), respectively by logistic regression analysis (denoted as c[ ]) are given in Figure 1 a-c, and Tables 3 and 4. In Table 3, the areas under the curves (AUC) are listed and compared to those of PSA-T alone or to the ratio PSA-F/PSA-T, respectively. All AUCs of the ratios as well as of the combinations obtained through logistic regression analysis were significantly improved over that of PSA-T. Comparing AUCs to that of PSA-F/PSA-T, only those given by the combinations of (hK2, PSA-F) and (hK2, PSA-F, PSA-T) by logistic regression analysis were significantly improved.

Comparison of the performance of each analysed parameter with that of PSA-T and PSA-F/PSA-T at given sensitivity and specificity levels are shown in Table 4. At a 90 percent sensitivity level, all the ratios analysed and logistic regression analysis combinations significanlty improved the specificity levels over that obtained by PSA-T. At this sensitivity level the specificity obtained by PSA-T (7%) was increased to 41% by hK2/PSA-F and to 63% by c[hK2,PSA- F] . Corresponding values for a sensitivity level of 80% were 23% (PSA-T), 63% (hK2/PSA-F) and 74% (c[hK2, PSA-F] ) . The (hK2/PSA-F)x(PSA-T/PSA-F) and c[hK2 , PSA-F, PSA-T] performed even somewhat better, an effect which was further accentuated at the 70 percent sensitivity level. On the other hand, when comparing the obtained specificities to that of the PSA-F/PSA-T ratio, the tested parameters resulting in significant specificity improvements are those containing hK2 (the two hK2 containing ratios and the two combinations through logistic regression analysis containing hK2). Using c[hK2,F] the following improvements were obtained relative to PSA-F/PSA-T: from 14 to 60% at 95%, from 27 to 63% at 90%, from 52 to 74% at 80% and from 67 to 80% at 70% sensitivity, respectively.

Analysing the ROC curves from the other end i.e. for sensitivity improvements at high specificity levels, it was observed that the analysed parameters containing hK2 and PSA-F generally give a significant improvement over PSA-T. For example at a 90 percent specificity level, the sensitivity of PSA-T was 16% but increased to 56% for hK2/PSA-F, to 64% for (hK2/PSA-F ) x(PSA-T/PSA-F ) and to 64% for c[hK2,F,T]. For a 80 percent specificity level corresponding percentages were, from 36 to 60%, to 76 and to 76% respectively. Compared to the sensitivites obtained by PSA-F/PSA-T, the ratio of (hK2/PSA-F ) x (PSA-T | /PSA-F ) and c[hK2, PSA-F, PSA-T] were the only parameters resulting in significant improvements (at 90 and 95% specificity levels ) .

When replacing PSA-T with PSA-ACT in the ROC analyses based on the use of ratios or combinations through logistic regression analysis, the performance was virtually unchanged (data not shown). Neither did the addition of

PSA-ACT to c[hK2, PSA-F, PSA-T] improve the performance (data not shown). This unequivocally showed that replacing PSA-T with PSA-ACT is fully feasible in maintaining but not significantly improving the differential diagnosis. Since PSA-T by the H117/H50 Mab combination detects the sum of PSA and hK2, we also checked whether correction of this concentration by subtraction of the hK2 concentration to give the "true" PSA-T concentration, would have any significant influence on the ROC analyses made: It had no significant effect.

In SP2, in which are included only patients with the total PSA range of 3 to 10 μg/L, hK2 correlated with high significance (p< 0.0001) with PSA-T, PSA-F and PSA-ACT in the BPH group, but with unimpressive correlation coefficients (r) 0.43, 0.56 and 0.29, respectively. For the PCa group, the corresponding correlation coefficients were 0.22, 0.66 and 0.05 respectively, with only the correlation between hK2 and PSA-F reaching statistical significance (p = 0.0003) .

Results SP3

Of the single parameters measured in SP3, the screening material, PSA-T and hK2 discriminated between cancers and non-cancers in a highly significant fashion (Table 5). So did the the PSA-F/PSA-T ratio and the two hK2 containing ratios which also included both PSA-F and PSA-T as components, whereas the ratios hK2/PSA-T (p=0.12) and hK2/PSA-F (p=0.056) did not. As shown in Table 6, restricting the analysis to the PSA-T range 3-10 μg/L, the diagnostic grey area in PSA diagnostics of PCa, following parameters become discriminative between PCa and benign cases: PSA-F, the PSA-F/PSA-T ratio and all hK2 ratios containing PSA-F as a component. PSA-T, hK2 and the ratio of hK2/PSA-F were unable to discriminate between cancer and non cancer cases .

The measured hK2 concentrations correlated significantly with the concentrations of PSA-T but with unimpressive correlation coefficients when all benign (r=0.336) or cancer (r=0.402) cases were included in the analysis. No significant association was seen between concentrations of hK2 and PSA-F. In the diagnostic grey area of PSA-T 3-10 μg/L, the significant association between hK2 and PSA-T was lost both for cancers and benign cases .

In Figure 2 a-c, the AUCs from ROC curves are shown and compared for the PSA-T 3-10 μg/L range of the SP3 material: the three single parameters, three ratios of which two involve hK2 , and three combinations by logistic regression analysis of (PSA-T, PSA-F, hK2 ) , (PSA-F, hK2 ) , and (PSA-F, PSA-T). Significant improvements of AUCs versus that of PSA-T were given by PSA-F/PSA-T, c[T,F] and c[hK2,F,T]. The ratio of (hK2/PSA-F ) x(PSA-T/PSA-F) gave a p-value just above the 0.05 limit. Relative to the AUC of PSA-F/PSA-T, no improvements were obtained .

In Table 8, diagnostic specificities and sensitivities from the ROC analyses (Fig. 2) are compared for the different parameters. As for the ratio PSA-F/PSA-T (and c[T,F]), statistically significant improvements (range +4 to +25 percents) in specificities at 90, 80 and 70 percent sensitivities were obtained for all parameters containing at least hK2 and PSA-F, in comparison to those obtained by total PSA alone. Similar statistically significant increases in sensitivities were obtained for specified high levels of specificity. Especially powerful here was the algorithm (hK2/PSA-F) x(PSA-T/PSA-F) , which gave increased specificities (range +19 to + 29 percent) relative to those of PSA-T at the 70 to 95 percent specificity levels. These were even improved (but did not reach statistical significance in this material) over the sensitivities obtained by PSA-F/PSA-T.

In Table 9 the results are shown from statistical comparisons of the cancers divided into two groups based on the pathological examination of the biopsy derived tissue specimens. The Gleason score is the best established predictor (prognostic marker) of a tumors^' pathological potential. Poorly differentiated tumors (high Gleason scores) are characterized by aggressive disease, well differentiated tumors (low Gleason scores) have indolent courses (Cookson et al . J Urol 1997; 157:559). Including all 80 cancers in the comparison, many of the analysed parameters significantly discriminated between the gleason 4-6 and Gleason 7-8 groups (most notably PSA-T, PSA-F/PSA- T, hK2 and hK2xPSA-T/PSA-F ) . In the clinically interesting PSA-T range of 3 to 10 μg/L (when the tumor^'s likelihood of being organ confined is still high) only the ratios hK2/PSA-F and (hK2/PSA-F) x(PSA-T/PSA-F) remained statistically significant discriminators of the two groups with hK2 and hK2xPSA-T/PSA-F showing positive trends. However in this PSA-T range, both PSA-T and PSA-F/PSA-T clearly lost their discriminatory capabilities.

DISCUSSION AND CONCLUSIONS

In this study we have established a methodological concept for a sensitive and specific immunometric determination of hK2 in the circulation. The low detection limit of the assay is achieved by sandwiching reagents of excellent technical performance (especially high affinity constants) in a carefully optimized assay design. The good specificity of the assay especially with regards to the crossreactivity with PSA was achieved predominantly due to a scavenger procedure wherein three PSA specific Mabs in excess are preventing PSA or PSA complexes from participating in the sandwiching (capture/detection) reaction. We have also established that this hK2 specific immunoassay behaves in a closely equimolar fashion towards hK2 complexes with two of its most prominent serpin inhibitors, ACT and PCI.

From this study it has been established that concentrations of hK2 in the studied populations do correlate with PSA-T and PSA-F but in a crude way and with unimpressive correlation coefficients. Restricting the analysis to the clinically interesting area of PSA-T below 10 (where the likelihood of finding a prostate cancer confined to the gland is still high) these associations become either weaker or disappear completely. These observations support the notion of hK2 as an independent analytical variable conceivably offering different clinical information compared to PSA-F and PSA-T.

Although it could be argued that hK2 on its own, like PSA, is an independent prostate cancer marker, the main conclusion from our studies is that the clinically most useful information of specific determinations of hK2 is to be found when it is combined with measurements of the forms of PSA and notably the free form of PSA. This is largely also a consequence of restricting the analysis to those patients still having a moderately increased PSA-T concentration, in which case a tumor is likely to be organ confined or localized and eligible for therapy interventions with a curative aim.

In the study population 2 (SP2), consisting of men with a median age of 70 to 71 years and presenting with clinical symptoms (due to PCa or BPH) and with total PSA between 3 and 10 μg/L, ROC analyses reveal PSA-T and hK2 as single parameters to be poor discriminators of PCa and BPH.

However, various ratios, notably that of hK2/PSA-F as such or in combination with PSA-T (multiplied) or PSA-F/PSA-T (divided), gave significant improvements in both diagnostic specificity at high sensitivity levels as well as in diagnostic sensitivities at high specificity levels not only in relation to PSA-T alone but also frequently and significantly in relation to the more recently introduced ratio PSA-F/PSA-T. Also interestingly, replacing the ROC analysis based on precalculated ratios of hK2, PSA-F (and PSA-T), by studying these same parameters in ROC analysis after combination through logistic regression analysis gave furthermore substantial improvements in the diagnostic performance.

In study population 3 (SP3), a screening population consisting of younger asymptomatic men, similar results were obtained with the important distinction that no significant improvements of the results given by PSA-F/PSA- T could be observed.

An important observation of the cancers identified in the screening procedure was seen in relation to the Gleason scores obtained from analyses of the biopsy specimens. In the total PSA range from 3 to 10 μg/L PSA-T, PSA-F and PSA F/PSA-T failed to discriminate between the Gleason score groups 4-6 and 7-8, whereas hK2/PSA-F and (hK2/PSA-F ) x(PSA- T/PSA-F) discriminated betweeen the groups significantly (p < 0.05). This observation supports the possibility that measurements of hK2 may provide a prognostic indicator of a tumors^' likelhood to behave aggressively in the future.

It will be appreciated that the methods of the present invention can be incorporated in the form of a variety of embodiments, only a few of which are disclosed herein. It will be apparent for the specialist in the field that other embodiments exist and do not depart from the spirit of the invention. Thus, the described embodiments are illustrative and should not be construed as restrictive.

Table 1. Study population 1 : Median, 25th, and 75th percentiie levels in serum of PSA- T, PSA-F, hK2, PSA-ACT, PSA-F/PSA-T, hK2/PSA-T, hK2/PSA-F, PSA-F/PSA-ACT, hK2^χPSA-T/PSA-F and (hK2/PSA-F)^χ(PSA-T/PSA-F) for patients with benign prostatic hyperplasia (BPH) and prostate cancer (PCa). Statistical significant differences between the groups are tested with the non-parametric Mann Whitney-U test (* denotes statistically significant i.e. p-value < 0.05).

BPH PCa p-values

(n=289) (n=104)

T, μg/L 4.3 (2.1 ; 7.0) 11.3 (6.0; 57.9) <0.0001 *

F, μg/L 0.86 (0.42; 1.5) 1.5 (0.72; 7.1) <0.0001 * hK2, μg/L 0.06 (0.03; 0.09) 0.18 (0.07; 0.97) <0.0001 *

PSA-ACT, μg/L 3.6 (1.8; 6.2) 10.2 (5.3; 43.6) <0.0001 ^*

F/T, % 22 (17; 28) 12 (8.6; 17) <0.0001 * hK2/T, % 1.5 (0.9; 2.3) 1.6 (0.9; 2.9) 0.25 hK2/F, % 6.9 (4.8; 10.2) 13 (8.1 ; 21.7) <0.0001 *

F/PSA-ACT, % 25 (18; 34) 13 (9.0; 19) <0.0001 * hK2xT/F, % 29 (14; 48) 157 (51 ; 772) <0.0001 *

( K2 F)x(T/F), % 32 (20; 48) 107 (62; 169) <0.0001 *

Table 2. Study population 2: Median, 25th, and 75th percentiie levels in serum of PSA- T, PSA-F, hK2, PSA-ACT, PSA-F/PSA-T, hK2/PSA-T, hK2/PSA-F, PSA-F/PSA-ACT, hK2^χPSA-T/PSA-F and (hK2/PSA-F)^χ(PSA-T/PSA-F) for patients with benign prostatic hyperplasia (BPH) and prostate cancer (PCa) within the total PSA range 3-10 μg/L. Statistical significant differences between the groups are tested with the non-parametric Mann Whitney-U test (* denotes statistically significant i.e. p-value < 0.05).

BPH PCa p-values

(n=90) (n=25)

T, μg/L 5.6 (4.2; 6.6) 5.7 (4.3; 7.7) 0.48

F, μg/L 1.1 (0.79; 1.4) 0.72 (0.59; 1.1) 0.004 * hK2, μg/L 0.07 (0.05; 0.1) 0.09 (0.06; 0.15) 0.04 *

PSA-ACT, μg/L 4.6 (3.5; 6.2) 4.9 (3.7; 7.2) 0.35

F/T, % 21 (15; 25) 11 (10; 18) 0.0004 * hK2/T, % 1 (1 ; 2) 2 (1 ; 3) 0.12 hK2/F, % 7 (5; 10) 12 (8; 18) <0.0001 ^*

F/PSA-ACT, % 24 (16; 32) 13 (10; 22) 0.0018 * hK2xT/F, % 36 (27; 52) 90 (44; 113) <0.0001 *

(hK2/F)x(T/F), % 36 (22; 52) 98 (60; 164) <0.0001 *

Table 3. Study population 2: Areas under the ROC curve (AUC) . PSA-T range 3-10 μg/L, n(PCa)=25, n(BPH)=90. Statistically significant improvements of AUC (p<0.05) compared to T or F/T are denoted with *.

Parameter Area SE P (ref. T) p (ref. F/T)

T 0.537 0.067 - 0.0209

F 0.700 0.056 0.0590 0.4342 hK2 0.630 0.067 0.2420 0.2504

F/T 0.732 0.059 0.0209 * hK2/F 0.792 0.054 0.0027 * 0.4298

(hl^/F)^χ(T/F) 0.831 0.049 _.0,0003 * 0,0537 c [T,F] 0.740 0.058 0.0089 * 0.7932 c [hK2,F] 0.863 0.041 <0.0001 ^* 0.0383 c [hK2,F,T] 0.875 0.042 <0.0001 * 0.0098 *

Table 4. Study population 2: Comparison of ROC curves at a given sensitivity or specificity level. PSA-T range 3-10 μg/L, n(PCa)=25, n(BPH)=90. Statistically significant improvements of sensitivity or specificity (p<0.05) compared to T or F/T are denoted with * and ⁺, respectively.

Specificity at given sensitivity level (%) Sensitivity at given specificity level (%)

Parameter 95 90 8Q ZO ZQ 80 9Q 95

T 6 7 23 30 40 36 16 4

F 17^* 19* 56* 59* 68 56 20 12 hK2 19 19 37 47 52 44 40 16

F/T 14 27* 52^* 67^* 64 56 36 24 hK2/F 37 ^*+ 41 * 63* 68* 64 60 56^* 48*

(hK2/F)^χ(T/F) 29 ^*+ 33* 76 *⁺ 86 *⁺ 84^* 76* 64 ^*+ 48 *⁺ c[T,F] 20^* 24* 50^* 67* 64 52 48^* 24 c[hK2,F] 60 ^*+ 63 *⁺ 74 *⁺ 80 *⁺ 80^* 76* 60^* 52* c[hK2,F,T] 46 *⁺ 61 *⁺ 76 ^*+ 89 *⁺ 84^* 76* 64 ^*+ 60 *⁺

Table 5. Study population 3: Median, 25th, and 75th percentiie levels in serum of PSA- T, PSA-F, PSA-F/PSA-T, hK2, hK2/PSA-T, hK2/PSA-F, hK2^χPSA-T/PSA-F and (hK2/PSA-F)^χ(PSA-T/PSA-F) for participants in the Gothenburg screening study with (cancer) and without (benign) prostate cancer. Statistical significant differences between the groups are tested with the non-parametric Mann Whitney-U test (^* denotes statistically significant i.e. p-value < 0.05).

Benign Cancer p-values (n=280) (n=80)

T, μg/L 4.1 (3.2; 6.1) 6.6 (4.0; 12.8) <0.0001 *

F, μg/L 0.80 (0.56; 1.2) 0.80 (0.56; 1.2) ?? 0.40 hK2, μg/L 0.03 (<0.03; 0.05) 0.04 (<0.03; 0.07) 0.0004 *

F/T, % 19 (15; 24) 11 (7.8; 16) <0.0001 * hK2/T, % 0.7 (0.5; 1.2) 0.7 (0.4; 1.0) 0.12 hK2/F, % 4.2 (2.6; 6.0) 4.9 (3.0; 9.3) 0.056 hK2xT/F, % 17 (11 ; 26) 31 (20; 70) <0.0001 *

(hK2/PSA-F)> <(T/F), % 21 (13; 38) 41 (23; 100) <0.0001 *

Table 6. Study population 3: Median, 25th, and 75th percentiie levels in serum of PSA-T, PSA-F, PSA-F/PSA-T, hK2, hK2/PSA-T, hK2/PSA-F, hK2^χPSA-T/PSA-F and (hK2/PSA-F)^χ(PSA-T/PSA-F) for participants in the Gothenburg screening study with (cancer) and without (benign) prostate cancer within the total PSA range 3-10 μg/L. Statistical significant differences between the groups are tested with the non-parametric Mann Whitney-U test (* denotes statistically significant i.e. p- value > 0.05).

Benign Cancer p-values (n=204) (n=41)

T, μg/L 4.5 (3.7; 6.0) 5.1 (3.9; 6.2) 0.35

F, μg/L 0.88 (0.63; 1.2) 0.64 (0.53; 0.98) 0.01 * hK2. μg/L 0.03 (<0.03; 0.05) 0.04 (<0.03; 0.06) 0.14

F/T, % 18 (14: 24) 15 (10; 18) 0.0002 * hK2/T, % 0.6 (0.5; 1.1) 0.6 (0.5; 1.1) 0.28 hK2/F, % 4.1 (2.6; 5.6) 5.7(3.6;8.4) 0.0026 * hK2xT/F, % 18 (12; 27) 26 (i 8; 41; 0.0014 *

(hK2/F)x(T/F), % 20 (13; 38) 44 (20; 87) 0.0003 *

Table 7. Study population 3: Areas under the ROC curve (AUC) within PSA-T range

3-10 μg/L, n(PCa)=41 , n(non-cancer)=204. Statistically significant improvements of

AUC (p<0.05) compared to T or F/T are denoted with *.

Parameter Area SE P (ref. T) p (ref. F/T)

T 0.546 0.049 - 0.04

F 0.627 0.048 0.24 0.40 hK2 0.571 0.049 0.73 0.091

F/T 0.683 0.046 0.04 ^* - hK2/F 0.649 0.049 0.14 0.61

(hK2/F)^χ(T/F) 0.680 0.050 0.056 0.96 c [T,F] 0.686 0.045 0.036 * 0.96 c [hK2,F] 0.629 0.049 0.23 0.42 c [hK2,F,T] 0.686 0.046 0.038 ^* 0.97

Table 8. Study population 3: Comparison of ROC curves at a given sensitivity or specificity level within PSA-T range 3-10 μg/L. n(PCa)=41, n(non-cancer)=204. Statistically significant improvements of sensitivity or specificity (p<0.05) compared to T or F/T are denoted with * and ⁺, respectively.

Parameter 95 90 80 ZO ZO 80 90 95

T 9 11 25 36 34 22 12 5

F 15 21 * 32 45 54 44* 15 10 hK2 0 0 0 43 42 27 10 7

F/T 8 28* 48* 60* 51 42^* 24 15 hK2/F 15 15* 25* 50^* 54 46^* 22 10

(hK2/F)^χ(T/F) 11 17 35* 57^* 58* 51 * 37^* 24* c[T,F] 5 26* 47* 62* 54* 39^* 24 17 c[hK2,F] 12 ⁺ 21 * 32^* 51 * 51 46* 14 10 c[hK2,F,T] 6 25* 47* 61 * 58* 39* 24 15

Table 9. Study population 3: Comparison of prostate cancer (PCa) patients with low (4-6) and high (7-8) Gleason grades. Statistically significant differences between patients with low and high Gleason scores are tested with the non-parametric Mann Whitney-U test (* denotes statistically significant i.e. p-value < 0.05).

p-values for PCa patients p-values for all PCa patients within PSA-T range 3-10 μg/mL Gleason: n(4-6)=63, n(7-8)=17 Gleason: n(4-6)=35, n(7-8)=6

T 0.0029 * 0.38

F 0.048 * 0.53

F T 0.0086 * 0.23

hK2 0.0004 * 0.083

hK2/T 0.51 0.13

hK2/F 0.078 0.046 *

hK2^χϊ/F 0.0009 * 0.060

(hK2/F)^χ(T/F) 0.016 * 0.027 ^*

Claims

1. A method for differentiating patients with cancer of the prostate (PCa) from patients with benign prostatic hyperplasia (BPH) and healthy male subjects without PCa, wherein the individual's body fluid concentration of human glandular kallikrein 2 (hK2) and prostate specific antigen (PSA) have been determined, characterized in that the combination of hK2 and PSA, wherein PSA means the free PSA, the complexed PSA or the total PSA, is used as marker distinguishing PCa patients from BPH and other non-PCa individuals.

2. The method according to claim 1, characterized in that the marker is

- the ratio hK2/F or the ratio F/hK2, wherein F is the concentration of free PSA, or - or the ratio hK2/F or the ratio F/hK2 multiplied by a quantity characterizing the PSA concentration.

3. The method according to claim 2, characterized in that the marker is hK2/F times T ( (hK2 x T)/F), wherein T is the concentration of total PSA.

4. The method according to claim 2, characterized in that the marker is hK2/F times T/F ((hK2/F) x T/F), wherein T is the concentration of total PSA.

5. The method according to claim 2, 3 or 4 , characterized in that the cut off limit for the ratio is chosen in different ways depending on whether high detection sensitivity or high detection specificity is preferred.

6. The method according to claim 1, characterized in that the combination of hK2 and PSA, wherein PSA means the free PSA, the complexed PSA or the total PSA, is a logistic regression analysis combination.

7. The method according to claim 6, characterized in that the logistic regression analysis combination is c[hK2, F], wherein F is free PSA.

8. The method according to claim 6, characterized in that the logistic regression analysis combination is c[hK2, F,

T], wherein F is free PSA and T is total PSA.

9. The method according to any of the foregoing claims, characterized in that the individual's body fluid is serum, plasma or whole blood.

10. The method according to any of the foregoing claims, characterized in that the individual's have total PSA in the range 1 to 20 ╬╝g/L.

11. The method according to claim 10, characterized in that the individual's have total PSA in the range 3 to 10 ╬╝g/L.

12. A method for differentiating patients with indolent or slowly progressive prostate cancers from patients with aggressively progressive prostate cancers, wherein the individual's body fluid concentration of human glandular kallikrein 2 (hK2) and prostate specific antigen (PSA) have been determined, characterized in that the combination of hK2 and PSA, wherein PSA means the free PSA, the complexed PSA or the total PSA, is used as marker distinguishing PCa patients with indolent or slowly progressive prostate cancers from patients with aggressively progressive prostate cancers.

13. The method according to claim 12, characterized in that the marker is

14. The method according to claim 13, characterized in that the marker is hK2/F times T/F ((hK2/F) x T/F), wherein T is the concentration of total PSA.

15. The method according to claim 13 or 14, characterized in that the cut off limit for the ratio is chosen in different ways depending on whether high detection sensitivity or high detection specificity is preferred.

16. The method according to claim 12, characterized in that the combination of hK2 and PSA, wherein PSA means the free PSA, the complexed PSA or the total PSA, is a logistic regression analysis combination.

17. The method according to claim 16, characterized in that the logistic regression analysis combination is c[hK2, F], wherein F is free PSA.

18. The method according to claim 16, characterized in that the logistic regression analysis combination is c[hK2, F, T], wherein F is free PSA and T is total PSA.

19. The method according to any of the claims 12 to 18, characterized in that the PCa-patients have total PSA in the range 1 to 20 ╬╝g/L, preferably in the range 3 to 10 ╬╝g/L.