Introduction. Objective To investigate the clinical significance of the 1.13 ng/ml, P<0.001) and a lower free-to-total PSA

Transcription

1 British Journal of Urology (1998), 81, Free-to-total prostate-specific antigen (PSA) ratio improves the specificity for detecting prostate cancer in patients with prostatism and intermediate PSA levels F. RECKER, M.K. KWIATKOWSKI, T. PIIRONEN*, K. PETTERSSON*, M. GOEPEL and R. TSCHOLL Clinic of Urology, Kantonsspital Aarau, Aarau, Switzerland and *Department of Biotechnology, University of Turku, Turku, Finland and Clinic of Urology, University of Essen Medical School, Essen, Germany Objective To investigate the clinical significance of the 1.13 ng/ml, P<0.001) and a lower free-to-total PSA free-to-total prostate-specific antigen (PSA) ratio in ratio (12.1% vs 19.9%, P<0.001) than those in group improving the specificity of PSA measurement for 2. The dicerences were similar between group 1 and detecting prostate cancer within the diagnostic intermediate group 3 (median free PSA in group 3, 1.06 ng/ml, range (4 10 ng/ml total PSA) in patients P=0.03, and free-to-total PSA ratio 18.7%, P= referred for the treatment of urinary symptoms ). There were no significant dicerences between Patients and methods Serum samples were obtained patients in groups 2 and 3. The free-to-total PSA ratio from 333 consecutive patients with obstructive and had a higher specificity than total PSA at all sensitivity irritative urinary symptoms. Of these men, 114 had levels, e.g. a threshold free-to-total PSA ratio of total PSA levels of 4 10 ng/ml; 22 had prostate detected 91% of cancers and spared 48% (group 2) or cancer (group 1) and 71 had benign prostatic hyper- 46% (group 3) from unnecessary biopsies. The area plasia (BPH, group 2). Group 3 consisted of 21 patients under the receiver operating characteristic curve for with BPH and a chronic indwelling catheter. The group 1 vs group 2 was 0.56 (total PSA) and 0.78 concentrations of free and total PSA (ProStatusTM, (free-to-total PSA ratio) and for group 1 vs group 3 Wallac Oy, Turku, Finland) and PSA complexed to was 0.56 (total PSA) and 0.81 (free-to-total PSA ratio). a-1-antichymotrypsin were measured and the free-tototal Conclusion In those patients with extensive symptoms PSA ratio calculated. All patients under 70 years from BPH and requiring surgical treatment, the free- of age or with suspicious findings on digital rectal to-total PSA ratio improves the specificity for detecting examination or transrectal ultrasonography under- prostate cancer in the diagnostic grey zone of went ultrasound-guided sextant prostate biopsies. Of 4 10 ng/ml total PSA. This improvement occurred the 114 patients, 105 (92%) underwent transurethral in patients with or without a chronic indwelling resection of the prostate and six (5%) radical retropubic catheter for urinary retention. prostatectomy. Keywords Prostate-specific antigen, free/total PSA, prostate Results Patients in group 1 had significantly lower cancer, transurethral catheter, benign prostatic median free PSA concentrations (0.78 ng/ml vs hyperplasia ing are clinically significant [6,7]. However, PSA level is Introduction tissue- and not tumour-specific; moreover, serum values There were over cases of prostate cancer and in patients with benign and malignant conditions overover cancer-related deaths in the USA in 1995 lap, and thus it does not fulfil the criteria for an ideal [1]. Since the introduction of PSA as a serum marker marker. In a screening population of men with total for prostate cancer [2] there has been a significant PSA levels in the grey zone (4 10 ng/ml) about 22% increase in cases of newly diagnosed prostate cancer. An have prostate cancer [8]. Thus, three patients undergo enormous ecort is put into early detection because only an unnecessary biopsy to detect one with cancer. patients with small, organ-confined cancers can be cured. DiCerent concepts have been used to enhance the There is ample evidence that screening with PSA more specificity of serum PSA measurement without losing often detects organ-confined cancers than a DRE alone sensitivity, including PSA velocity, density and age- [3 5] and that cancers detected using serum PSA screen- specific reference ranges [9 11]. Furthermore, PSA exists in serum in several forms, of which the most important Accepted for publication 16 December 1997 are the free form of PSA (fpsa), PSA complexed to British Journal of Urology

2 FREE-TO-TOTAL PSA FOR DETECTING PROSTATE CANCER 533 a-1-antichymotrypsin (PSA-ACT), and PSA complexed indications to operative treatment. The remaining to a-2-macroglobulin. The sum of PSA-ACT and fpsa patients underwent TURP. mainly accounts for the total PSA (tpsa) in the serum Blood samples were obtained before any prostatic [12,13]. The observation of significant dicerences in the manipulation. After clot formation, the samples were proportion of fpsa in patients with prostate cancer when centrifuged and serum collected and frozen at 70 C. compared with those with BPH or in healthy men The samples were thawed immediately before measure- [12 15] originated a new concept based on the ratio of ment. Total and fpsa were determined using the commercially serum fpsa to serum tpsa, the free-to-total PSA ratio available monoclonal immunofluorometric assay (f/tpsa). Recently, a better discrimination between pros- (ProStatusTM PSA Free/Total, Delfia reagents, Wallac Oy, tate cancer and BPH was reported using f/tpsa rather Turku, Finland) [20]. PSA-ACT was also measured using than tpsa, within the diagnostic grey zone of tpsa a fluorescence immunoassay [11,21] in 45 of 72 patients levels of 4 10 ng/ml [14,16 19]. However, the study in group 2, 16 of 21 in group 3 and 11 of 22 in group 1. populations mostly consisted of mixed or screened partici- Descriptive statistics were calculated for each study pants. We have investigated the usefulness of f/tpsa in subgroup, using the nonparametric Kruskal Wallis (chi- a selected group of patients with extensive symptoms square) test to determine the significance of dicerences of BPH. in the median values of total, free and f/tpsa, as well as other variables between groups. For all analyses, P<0.05 was considered to indicate statistical significance. Across Patients and methods the range of tpsa investigated, the sensitivity and The study population consisted of 333 consecutive specificity for tpsa and f/tpsa were calculated using patients with LUTS admitted to our department between ROC curves, generated by plotting sensitivity against February 1996 and January Of these, 219 were 1-specificity. The areas under the curve (AUC) were then admitted with symptomatic BPH, the mean (sd) IPSS calculated [22]. for uncatheterized patients being 20.5 (7.0), with an obstructive score of 11.9 (4.7) and an irritative score of 8.6 (3.2); 69 had prostate cancer and 45 had BPH Results requiring an indwelling Foley catheter before admission. The distribution of patient age at diagnosis, tpsa, fpsa, From this population, a group of 114 patients were f/t PSA and PSA-ACT are presented in Table 1 for all selected who had a tpsa of 4 10 ng/ml (the diagnostic 333 patients, and for the subgroup of 114 patients with grey zone ). The inclusion criteria were: no history of a tpsa of 4 10 ng/ml, with descriptive statistics for the previous treatment of prostatic disease, no previous subgroups. In the diagnostic grey zone, the median prostatic manipulation within 3 weeks and no signs of tpsa and PSA-ACT did not dicer significantly among infection. These 114 patients were divided into three the subgroups but the fpsa and f/tpsa dicered between groups comprising: group 1, 22 patients with prostate group 1 and groups 2 and 3. There were no significant cancer but with no catheter; group 2, 71 patients with dicerences between groups 2 and 3, and no significant BPH with no catheter; and group 3, 21 patients with correlation of f/tpsa with tpsa level (Fig. 1). The ratio BPH and an indwelling catheter. The patients in group of PSA-ACT to tpsa did not dicer significantly among 3 had had the catheter for 9 23 days before operation. the groups (data not shown). PSA-ACT was confirmed In this group, the pathological findings of the surgical to be the major complex of immunodetectable PSA but specimens showed acute severe inflammation in three, the ratio PSA-ACT to tpsa failed to improve the speci- mild chronic inflammation in 14, chronic severe ficity for detecting cancer when compared with tpsa. inflammation in one and no inflammation in three. A For all patients with no indwelling catheter (69 with DRE was performed in all patients; as a clinical routine, cancer and 219 with BPH, Table 1), the sensitivity, all patients younger than 70 years, or with a minimal specificity and positive predictive value (PPV) at a tpsa life expectancy of 10 years or with a suspicious DRE threshold of 4 ng/ml were 88%, 57% and 40%, respectively; underwent TRUS-guided sextant prostate biopsies. In hence, at this threshold, eight of 69 patients with those patients with a tpsa level of 4 10 ng/ml, the cancer would be undetected and 93 of 154 patients biopsy revealed cancer in 13 and the histological examination would have an unnecessary biopsy. For the f/tpsa, the of the surgical specimen detected an incidental same sensitivity (88%) was reached when a threshold of carcinoma in nine (four pt1a and five pt1b). Of the 13 was used, but the specificity and PPV were higher, cancers detected on biopsy, six patients underwent rad- at 66% and 45%, respectively. With a further decrease ical prostatectomy (five pt3, one pt2c), four a palliative of the threshold to 0.15, the sensitivity also decreased TURP (one ct2, two ct3 and one ct4), and three (all (70%) while the specificity (82%) and PPV (55%) ct2) were treated conservatively because of contra- improved.

3 534 F. RECKER et al. Table 1 Frequency distribution [number (%)] of age, tpsa and f/tpsa levels for all 333 patients (for the complete range of tpsa), and the distribution and statistical values for the subgroups in the intermediate range Prostate cancer (group 1) BPH (group 2) BPH+catheter (group 3) Total population (n) Age (years) (12) 25 (11) 2 (5) (26) 62 (28) 18 (40) (46) 95 (43) 15 (33) >80 11 (16) 39 (18) 10 (22) tpsa (ng/ml) <2 1 (1) 73 (33) 4 (9) (10) 53 (24) 8 (18) (32) 71 (32) 21 (47) (9) 16 (8) 10 (22) >20 33 (48) 6 (3) 2 (4) f/tpsa (%) <10 32 (46) 10 (4) 2 (4) (23) 29 (13) 11 (24) (19) 35 (16) 12 (27) (7) 41 (19) 9 (20) (2) 41 (19) 8 (18) >30 2 (3) 63 (29) 3 (7) Subgroups (tpsa ng/ml): Number Age (years) (14) 13 (18) 1 (5) (32) 19 (27) 12 (57) (40) 30 (42) 5 (24) >80 3 (14) 9 (13) 3 (14) Median (range) 71 (57 85) 71 (55 95) 70 (59 84) Mean (sd) 69.9 (7.4) 71.2 (9.0) 71.5 (7.1) tpsa (ng/ml) (41) 38 (53) 12 (57) (32) 19 (27) 6 (29) (27) 14 (20) 3 (14) Median (range) 7.17 ( ) 5.74 ( ) 5.57 ( ) Mean (sd) 6.72 (1.99) 6.23 (1.71) 5.96 (1.61) fpsa (ng/ml) 1 17 (77) 23 (32) 9 (43) (23) 39 (55) 9 (43) (0) 7 (10) 3 (14) (0) 2 (3) 0 (0) Median (range) 0.78 ( ) 1.13 ( ) 1.06 ( )* Mean (sd) 0.87 () 1.31 (0.67) 1.23 (0.57) f/tpsa (%) 10 7 (32) 5 (7) 0 (0) (36) 13 (18) 6 (28) (22) 19 (27) 5 (24) (5) 17 (24) 5 (24) (5) 8 (11) 4 (19) >30 0 (0) 9 (13) 1 (5) Median (range) 12.1 ( ) 19.9 ( ) 18.7 ( ) Mean (sd) 13.2 (5.3) 21.0 (8.8) 20.3 (6.7) PSA-ACT (ng/ml) <4 3 (28) 5 (11) 3 (19) (18) 19 (42) 7 (44) (36) 15 (34) 4 (25) (18) 6 (13) 2 (12) Median (range) 7.24 ( ) 5.58 ( ) 5.34 ( ) Mean (sd) 6.84 (3.26) 5.53 (1.80) 6.04 (2.15) P value of the dicerences between group 1 and 2 or 3, *<0.05 and <0.001; there were no statistically significant dicerences between groups 2 and 3.

4 FREE-TO-TOTAL PSA FOR DETECTING PROSTATE CANCER Free-to-total PSA ratio Fig. 1. The f/tpsa levels in group 1 (prostate cancer), 2 (BPH) and 3 (BPH plus catheter) plotted against tpsa levels. The bars show the median f/tpsa level Group Group 2 Total PSA (ng/ml) 4 10 Group 3 Within the grey zone of tpsa, the sensitivity and with complete specificity and sensitivity the area equals specificity for dicerent tpsa and f/tpsa thresholds for 1.0. The sensitivity was 91% at an f/tpsa threshold of groups 1 and 2 are shown in Fig. 2. From the graph, (Fig. 3b), with a specificity and PPV of 48% and dicerent thresholds can be selected depending on the 35%, respectively. Using this threshold would have compromise between sensitivity (detected cancers) and resulted in a negative test in 37 of 71 (52%) of patients specificity (unnecessary biopsies). with BPH. The sensitivity and specificity for group 1 vs Figure 3 summarizes these calculations by plotting the group 3 for dicerent f/tpsa thresholds are shown in true positive (sensitivity) vs false positive (1 specificity) Fig. 4. The AUC of the ROC curves plotted for these results on ROC curves, providing a graphical comparison groups is 0.56 for tpsa and 0.81 for f/tpsa (data not of test performance. The areas under the curves (AUCs) shown). At the f/tpsa threshold of, the sensitivity are directly proportional to the reliability of a given test; was 90%, the specificity 45% and the PPV 65%. Operating point on the ROC-curve a Specificity Sensitivity Total PSA threshold value 10 Operating point on the ROC-curve 0.00 b Sensitivity Specificity f/t PSA threshold value Fig. 2. A comparison of group 1 (prostate cancer) and 2 (BPH, no catheter), showing the sensitivity (red) and specificity (green) levels at dicerent, a, tpsa and, b, f/tpsa thresholds over the range of tpsa. The arrows in b show the specificity and f/tpsa thresholds at the 95% sensitivity level. The compromise between sensitivity (percentage of detected cancers) and specificity (percentage of spared biopsies) at all thresholds can be calculated.

5 536 F. RECKER et al. True positive fraction (sensitivity) f/t Ratio PSA Total 0.00 False positive fraction (1-specificity) Fig. 3. ROC curves comparing the performance of tpsa (red) and f/tpsa (green) in dicerentiating between patients with prostate cancer or BPH. The f/tpsa improves specificity for detecting cancer at the same sensitivity level compared with tpsa. The area under the curve was 0.78 for f/tpsa and 0.56 for tpsa. Operating point on the ROC-curve Sensitivity Specificity f/t PSA threshold value Fig. 4. A comparison of group 1 (prostate cancer) and 3 (BPH, with catheter), showing the sensitivity (green) and specificity (red) levels at dicerent f/tpsa thresholds. The arrows show the specificity and f/tpsa thresholds at the 95% sensitivity level. Interestingly, none of the patients in group 3 had a f/tpsa of<10%. Discussion The measurement of serum PSA has undoubtedly advanced the early diagnosis of clinically significant prostate cancer [2 4]. However, despite its high sensitivity, the specificity of this marker for cancer is relatively low. Since it was established that PSA exists in dicerent forms in serum, and that there are dicerences in the proportion of these forms in patients with cancer and those with benign disease [12 15], several investigators have tried to improve the detection of prostate cancer. In the present patients with prostatism, a threshold f/tpsa of gave 91% sensitivity and 48% specificity, higher than the values reported in the screened population by Catalona et al. [23], where the specificity was 38% at a threshold of 3, and of Reissigl et al. [24], with a specificity of 36% at a threshold of (but a similar sensitivity of 90%). In another screening population assessed by Bangma et al. [25], the sensitivity was 91% and the specificity 36% using a threshold f/tpsa of In a report where the selection of participants was similar to that in the present study (at least some of the patients with prostatism had to be treated surgically) van Cangh et al. [26] reported a sensitivity of 88% and specificity of 44% at an f/tpsa threshold of 0.235, for a tpsa of ng/ml. The higher specificity in the present symptomatic patients may be due to their higher prevalence of BPH compared with a younger screening population. Despite negative findings on DRE, TRUS and prostate biopsies in the present patients, a few low-volume cancers in the peripheral zone may have been undetected in those >70 years old. On the other hand, other studies in which the diagnosis was confirmed by sextant biopsy (with no additional transition zone biopsy) also failed to detect cancers originating in the transition zone, which were detected in the present study. The prevalence of cancer in the present selected patients was 22 of 114 (19%), comparable with the 21.7% reported by Catalona et al. [8] or the 23.2% recently reported by Bangma et al. [25]. To provide information for making clinical decisions, it might be helpful to screen the symptomatic and asymptomatic groups separately, for the policy of investigating all consecutive patients independently of the cause of referral results in drawing conclusions based on mixed populations, i.e. clinically undefined consecutive patients examined for prostate disease. To our knowledge, there are no previously published reports comparing f/tpsa between patients with cancer and those with BPH and an indwelling catheter, within the 4 10 ng/ml tpsa range. The present specificities and sensitivities for cancer were very similar for both groups with BPH and suggest that f/tpsa discriminates equally well for cancer between patients with and without an indwelling catheter. This may be clinically useful because an indwelling catheter can increase the serum tpsa level [27]. From the present results, a f/tpsa of < strongly suggests the presence of cancer in patients with an indwelling catheter. In the group of six symptomatic patients who under- went radical prostatectomy, five had stage pt3 tumours and only one was organ-confined. Although there were

6 FREE-TO-TOTAL PSA FOR DETECTING PROSTATE CANCER 537 few such patients, it seems that symptomatic patients 9 Carter HB, Pearson JD, Metter EJ et al. Longitudinal have a higher frequency of locally advanced disease than evaluation of prostate-specific antigen levels in men with asymptomatic screened patients. This is in agreement and without prostate disease. JAMA 1992; 267: with data reported by Schmid et al. [28] where only half 10 Benson MC, Whang IS, Olsson CA, McMahon DJ, Cooner WH. The use of prostate specific antigen density to enhance of detected cancers were organ-confined in a symptothe predictive value of intermediate levels of serum prostate matic population with tpsa levels of 4 10 ng/ml, and specific antigen. J Urol 1992; 147: supports the need for a separate analysis of symptomatic 11 Oesterling JE, Jacobsen SJ, Klee GG et al. Free, complexed and asymptomatic patients. and total serum prostate specific antigen: the establishment In conclusion, f/tpsa is clinically relevant in diceren- of appropriate reference ranges for their concentrations tiating prostate cancer from BPH in symptomatic patients and ratios. J Urol 1995; 154: referred to a urological centre. The specificity of f/tpsa 12 Stenman UH, Leinonen J, Alfthan H, Rannikko S, was higher than that for tpsa at all sensitivity levels Tuhkanen K, Alfthan O. A complex between prostate- and the overall advantage was apparent on ROC specific antigen and alpha 1-antichymotrypsin is the major analysis. Interestingly, similar results were obtained for form of prostate-specific antigen in serum of patients with patients with BPH and an indwelling catheter. The prostatic cancer: assay of the complex improves clinical sensitivity for cancer. Cancer Res 1991; 51: results suggest that the presence or absence of symptoms 13 Lilja H, Christensson A, Dahlen U et al. Prostate-specific of prostatism is an important aspect in defining the antigen in serum occurs predominantly in complex with population investigated. This approach may be useful alpha 1-antichymotrypsin. Clin Chem 1991; 37: for making decisions in daily practice. 14 Christensson A, Bjork T, Nilsson O et al. Serum prostate specific antigen complexed to alpha 1-antichymotrypsin as an indicator of prostate cancer. J Urol 1993; 150: Acknowledgements 15 Leinonen J, Lovgren T, Vornanen T, Stenman UH. We thank Dr B. Seifert, Institute of Biostatistics, Double-label time-resolved immunofluorometric assay of University of Zurich, for providing statistical help. prostate-specific antigen and of its complex with alpha 1-antichymotrypsin. Clin Chem 1993; 39: Partin AW, Catalona WJ, Southwick PC, Subong EN, Gasior References GH, Chan DW. Analysis of percent free prostate-specific antigen (PSA) for prostate cancer detection: influence of 1 Parker SL, Tong T, Bolden S, Wingo PA. Cancer statistics, total PSA, prostate volume, and age. Urology 1996; CA Cancer J Clin 1996; 46: : Stamey TA, Yang N, Hay AR, McNeal JE, Freiha FS, 17 Vashi AR, Wojno KJ, Henricks W et al. Determination of Redwine E. Prostate-specific antigen as a serum marker for the reflex range and appropriate cutpoints for percent free adenocarcinoma of the prostate. N Engl J Med 1987; prostate-specific antigen in 413 men referred for prostatic 317: evaluation using the AxSYM system. Urology 1997; 3 Labrie F, Dupont A, Suburu R et al. Serum prostate specific 49: antigen as pre-screening test for prostate cancer. J Urol 18 Prestigiacomo AF, Stamey TA. Can free and total prostate 1992; 147: specific antigen and prostatic volume distinguish between 4 Ellis WJ, Chetner MP, Preston SD, Brawer MK. Diagnosis men with negative and positive systematic ultrasound of prostatic carcinoma: the yield of serum prostate specific guided prostate biopsies? J Urol 1997; 157: antigen, digital rectal examination and transrectal ultrason- 19 Thiel RP, Oesterling JE, Wojno KJ et al. Multicenter ography. J Urol 1994; 152: comparison of the diagnostic performance of free prostate- 5 Catalona WJ, Smith DS, RatliC TL, Basler JW. Detection specific antigen. Urology 1996; 48: of organ-confined prostate cancer is increased through 20 Mitrunen K, Pettersson K, Piironen T, Bjork T, Lilja H, prostate-specific antigen-based screening. JAMA 1993; Lovgren T. Dual-label one-step immunoassay for simul- 270: taneous measurement of free and total prostate-specific 6 Smith DS, Catalona WJ. The nature of prostate cancer antigen concentrations and ratios in serum. Clin Chem detected through prostate specific antigen based screening. 1995; 41: J Urol 1994; 152: Pettersson K, Piironen T, Seppala M et al. Free and 7 Humphrey PA, Keetch DW, Smith DS, Shepherd DL, complexed prostate-specific antigen (PSA): in vitro stability, Catalona WJ. Prospective characterization of pathological epitope map, and development of immunofluorometric features of prostatic carcinomas detected via serum prostate assays for specific and sensitive detection of free PSA and specific antigen based screening. J Urol 1996; 155: PSA-alpha 1-antichymotrypsin complex. Clin Chem 1995; 8 Catalona WJ, Richie JP, Ahmann FR et al. Comparison of 41: digital rectal examination and serum prostate specific 22 Hanley JA, McNeil BJ. The meaning and use of the area antigen in the early detection of prostate cancer: results of under a receiver operating characteristic (ROC) curve. a multicenter clinical trial of 6630 men. J Urol 1994; Radiology 1982; 143: : Catalona WJ, Smith DS, Wolfert RL et al. Evaluation of

7 538 F. RECKER et al. percentage of free serum prostate-specific antigen to serum prostate-specific antigen level in benign prostatic improve specificity of prostate cancer screening. JAMA hyperplasia. Urology 1997; 49: ; 274: Schmid HP, Ravery V, Billebaud T et al. Early detection of 24 Reissigl A, Klocker H, Pointner J et al. Usefulness of the prostate cancer in men with prostatism and intermediate ratio free/total prostate-specific antigen in addition to total prostate-specific antigen levels. Urology 1996; 47: PSA levels in prostate cancer screening. Urology 1996; 48: Bangma CH, Rietbergen JB, Kranse R, Blijenberg BG, Petterson K, Schroder FH. The free-to-total prostate specific antigen ratio improves the specificity of prostate specific Authors antigen in screening for prostate cancer in the general F. Recker, MD, Deputy Head. population. J Urol 1997; 157: M.K. Kwiatkowski, MD, Resident. 26 Van Cangh PJ, De Nayer P, De Vischer L et al. Free to T. Piironen, MSc, Research Scientist. total prostate-specific antigen (PSA) ratio improves the K. Pettersson, PhD, Senior Scientist. discrimination between prostate cancer and benign pros- M. Goepel, MD, Senior Registrar. tatic hyperplasia (BPH) in the diagnostic gray zone of R. Tscholl, MD, Professor and Head of Clinic of Urology ng/ml total PSA. Urology 1996; 48: Correspondence: Dr med. Franz Recker, Clinic of Urology, 27 Batislam E, Arik AI, Karakoc A, Uygur MC, Germiyanoglu RC, Erol D. ECect of transurethral indwelling catheter on Department of Surgery, Kantonsspital Aarau, CH-5001 Aarau, Switzerland.