ADULT UROLOGY PROSPECTIVE EVALUATION OF PROSTATE-SPECIFIC ANTIGEN DENSITY AND SYSTEMATIC BIOPSIES FOR EARLY DETECTION OF PROSTATIC CARCINOMA MICHEL BAZINET, M.D. ALAA W. MESHREE M.D. CLAUDE TRUDEL, M.D. SAMUEL ARONSON, M.D. FRANCOIS Pt~LOQUIN, M.D. MAHMOUD NACHABE, M.D. LOUIS R. BI~GtN, M.D. MOSTAFA M. ELHILALI, M.D. From the Departments of Urology and Pathology, McGill University, Montreal, Quebec, Canada ABSTRACT--Significant controversies persist in regard to the need for systematic biopsies in patients with serum prostate-specific antigen (PSA) levels above 4 ng/mL (Hybritech assay), especially if they show no signs of prostatic cancer on digital rectal examination (DRE) or transrectal ultrasonography (TRUS). We evaluated 565 consecutive patients referred to us for prostatism, suspicious lesions on DRE, or an elevated serum PSA level. These patients do not represent a purely screened population. A detection rate of 38.4 percent was achieved by performing directed biopsies of suspicious lesions on DRE and/or TRUS, and systematic biopsies of all patients with serum PSA levels above 4 ng/mL. Among 142 patients with serum PSA between 4.1 and 10 ng/mL, but without suspicion for cancer on DRE and TRUS (DRE- TRUS-), a large number of patients (6.2) were subjected to systematic biopsies to detect one cancer. A receiver-operating characteristic curve for PSA density (PSAD) applied to this population confirmed that the best cut-off point for biopsies was a PSAD of 0.15, below which only two of twenty-three cancers would have been missed, sparing biopsies in 77 of 142 patients. A similar approach was applied to DRE- TRUS- patients with serum PSA levels above 10 ng/mL. The number of cancers in those with serum PSA between 10.1 and 14 ng/mL was too low to establish a PSAD cut-off point. In patients with serum PSA above 14 ng/mL, the best PSAD cut-off point for biopsies was 0.3, below which two of thirteen cancers would have been missed, sparing biopsies in 19 of 39 patients. We conclude that PSAD can safely reduce the number of patients subjected to systematic biopsies without significantly compromising cancer detection. The usefulness of mass screening for prostatic carcinoma by any method is an issue of debate and controversy. Improved prostate cancer detection among men who seek medical care is possible by the rational use of digital rectal examination (DRE) and prostate-specific antigen (PSA) measurement with transrectal ultrasonography of the prostate (TRUS) used in selected cases. 1 Serum PSA is superior to DRE in its ability to detect the possible presence of prostatic canSubmitted: April 28, 1993, accepted (with revisions): July 21, 1993 44 cer. 1,2 Since serum PSA levels can be increased by benign prostatic hyperplasia, 3 researchers described the use of PSA density (PSAD = serum PSA/prostate volume) as a potentially better predictor of prostate cancer. 4 Hodge et al. 5 reported that six systematic biopsies of the prostate, using transrectal ultrasound guidance, could increase the detection rate of prostatic cancer, provide better estimation of tumor volume and location, and allow more accurate representation of the average Gleason grade of the cancer. Theoretically, however, systematic biopsies may detect clinically insignificant cancers, can still UROLOGY / JANUARY1994 / VOLUME43, NUMBER1 miss larger cancers and apply mainly to peripheral and central zone cancers. 5 Apparently, none of the presently used tests is a perfect diagnostic procedure on its own. After studying 2,634 patients, Cooner 6 constructed an algorithm to serve as a practical guide for prostate cancer detection in patients seeking urologic care. He recommended performing six systematic biopsies in patients with serum PSA levels between 4.1 and 10 ng/mL only if their PSAD was _>0.15. Catalona, 7 however, recently suggested that all patients with s e r u m PSA levels above 4.0 n g / m L should be subjected to systematic biopsies. To resolve this controversy, we prospectively performed six systematic biopsies on 491 consecutive patients referred to us with serum PSA levels above 4 ng/mL, MATERIAL AND METHODS Between April 21 and November 9, 1992, we examined 565 consecutive patients aged forty-one to eighty-four years, by DRE, TRUS, and serum PSA determination by the Hybritech assay (normal range 0-4 ng/mL). All patients were referred to our center by their urologists for TRUS as part of their clinical evaluation. The reasons for referral were prostatic symptoms, suspicious DRE, or elevated PSA. It should be emphasized that these patients entered via a urologic practice and do not represent a purely screened population. Patients previously d i a g n o s e d to have prostatic cancer were excluded from the stud3a Serum PSA was k n o w n before performing the TRUS in all cases. DRE was systematically performed by the attending urologist just prior to TRUS with the patient in the left lateral decubitus position. The results were classified as either normal or suspicious. The prostate was considered to be suspicious if it was diffusely hard, contained a discrete firm to hard area, showed irregular contours or prominent lobe asymmetry. All patients were prepared by Fleet enema a few hours prior to TRUS. In addition, each of them r e c e i v e d 400 m g n o r f l o x a c i n or 500 m g ciprofloxacin twelve hours and one h o u r before examination. Those who underwent biopsies continued the antibiotic for two additional days. All TRUS examinations were performed in real time, using a Bruel and Kjaer model 1846 scanner with a 8551 m u l t i p l a n e probe (7 MHz transducer). Each gland was examined in both axial and sagittal projections. Patients with palpable abnormalities were carefully examined by TRUS for corresponding lesions. Prostate volume was calculated as follows: volume equals 0.52 x (length x width UROLOGY / JANUARY1994 / VOLUME 43, NUMBER 1 x height), with length being measured in the long i t u d i n a l view, a n d w i d t h a n d h e i g h t in the transaxial view. PSAD, a term suggested by Benson e t al., 4 was c a l c u l a t e d by d i v i d i n g PSA ( m o n o c l o n a l , expressed in ng/mL) by prostate volume (expressed in cubic centimeters). Final TRUS impressions were classified as benign or suspicious for cancer. The findings were considered to be suspicious if a hypoechoic lesion 8 seen in the prostate could not be explained by other factors (such as n o r m a l anatomic areas, vascular structures, cysts, or artifacts). 1 Biopsies were taken with the Bard spring-driven Biopty gun, using 18-gauge "Tru-cut" type biopsy needles. All biopsies were performed under ultras o u n d guidance via the transrectal route on an outpatient basis. In patients with serum PSA levels < 4 ng/mL, ultrasound-guided biopsies were directed toward suspicious areas on the ultrasound image, or toward suspicious areas detected by DRE by geographic c o o r d i n a t i o n (directed biopsies). Systematic biopsies were not done in this group of patients. Any patient with a s e r u m PSA level above 4 ng/mL was subjected to six systematic biopsies, with or without directed biopsies. The biopsies were equidistant, three for each lobe: at the apex, middle, and base of the prostate. With respect to the coronal plane, the biopsy sites were oriented in the center of each lobe, equidistant from the midline of the prostate and the lateral border of the lobe. The six cores were promptly placed in separate bottles containing 10% buffered formalin and were labeled accordingly to denote the six different sites. If a suspicious lesion was seen on DRE and/or TRUS, one to three directed biopsies were taken from the lesion, and the core(s) were i n s e r t e d in the b o t t l e of the c o r r e s p o n d i n g a n a t o m i c region. The cores were stained with h e m a t o x y l i n and eosin. The same p a t h o l o g i s t (L.R.B.) examined all slides. Each biopsy speci~ men was categorized histologically (normal tissue, atrophic changes, acute or chronic inflammation, prostatic intraepithelial neoplasia, or cancer), and all cancers were graded according to the Gleason scheme as a histologic score on a scale from 2 to 10. The percentage of cancer present in each core was estimated within a 5 percent range, The unpaired Student's t test was used (calculating 2-tail p values) to compare two means. The sensitivity and specificity of the diagnostic tests were calculated as described elsewhere. 9 The cutoff points were considered to be arbitrary values separating positive from negative test results. The 45 ....... PSAD (0.813) ~ PSA (0.728) TABLE I. I I lOO No. of patients No. of patients bit+psied DRE/'RUS - _<4 4.1 - 10 > 10 3i 142 89 0 142 89 23 20 (16.2) (22.5) 6.2 4.5 DRE + I"RUS - -< 4 4.1 - 10 > 10 14 35 25 14 35 ~5 1 12 15 (7.1) (34,3) (60,0) 14.0 2.9 1.7 DREI'RUS + ~4 4.1 - 10 > 10 8 33 36 8 33 36 1 10 19 (12.5) (313.3) (52.8) 8.0 3.3 1.9 ORE + rRUS + 4 4.1 - I0 > 10 21 54 77 21 54 77 6 38 72 (28.6) (70.4) (93.5) 3.5 1,4 1.1 70. g so ,m .>50, 4o. ¢/1 3020100 ,i 0 ..... 10 i,,,'/'l,,,',i,,,il 20 .... 'i",i ,, i, , ,', i . . . . 30 40 50 60 70 80 False Positive Fraction (%) I 90 (%) No. of patients biopsied to detect 1 with cancer .... 100 FIGURE 1. ROC curve comparing PSAD and serum PSA in all patients studied. Area under the curve is given in parentheses with a perfect test having an area of 1.0. receiver-operator characteristic (ROC) curve is a graphic plot of "sensitivity (%)" (true positive fraction) on the Y axis versus "1 - specificity (%)" (false-positive fraction) on the X axis at various cut-off points. The better a test is, the closer is its ROC curve to the upper left-hand corner of the graph, where the true-positive rate is 1 and the false-positive rate is 0. 9 To compare the Usefulness of two different diagnostic tests for the same target disorder, the area u n d e r the corresponding ROC curves is calculated and eomparedP We used the Wilcoxon statistic as described by Hanley and McNeil 1° to estimate the area u n d e r the ROC curve together with its standard error, and we assessed whether the difference in areas under two curves is random or real by calculating a Critical ratio z. u Since the prevalence of prostatic cancer in this population was not known, it was extrapolated from the biopsy results. RESULTS Of the 565 patients examined, biopsies were performed in 534 subjects, with a biopsy rate of 94.5 percent. However, all 491 patients with a serum PSA level above 4 ng/mL were subjected to six systematic biopsies (with or without directed biopsies) for a biopsy rate of 100 percent. Carcb noma was found in 217 cases, yielding a detection rate of 38.4 percent, and a positive biopsy rate of 40,6 percent. 46 No. of cancers detected PSA (ng/ml) 9o. 80. Overall results I A ROC curve was produced from the raw data on all patients, for both PSA and PSAD (Fig. 1). The areas under both curves were calculated, 1° being 72.8 percent for serum PSA and 81.3 percent for PSAD, with a statistically significant difference between both areas (p < 0.0001). The patients were divided into four main groups: the first group without suspected cancer on both DRE and TRUS (DRE- TRUS-); the second group with suspected cancer on DRE but w i t h o u t suspicious findings on TRUS (DRE+ TRUS-); the third group without suspected cancer on DRE b u t w i t h s u s p e c t e d c a n c e r on TRUS (DRE- TRUS+); and the last group with suspected cancer on both DRE and TRUS (DRE+ TRUS+). Each group was further divided into three subgroups according to their serum PSA level: the first with serum PSA ___4 ng/mL; the second with serum PSA between 4.1 and 10 ng/mL; and the last subgroup with serum PSA > 10 ng/mL. The overall results are summarized in Table I, DRE- TRUS- This group was comprised of 262 patients, of w h o m 31 had a serum PSA level N 4 ng/mL and were not subjected to biopsies, The others underwent systematic biopsies: twenty-three cancers were detected in 142 patients with serum PSA between 4.1 and i0 ng/mL, and twenty cancers were found in 89 subjects with serum PSA above 10 ng/mL. ROC curves were produced from the raw data on patientg with serum PSA levels between 4.1 and 10 ng/r~L (Fig. 2), above 10 ng/mL (Fig. 3), and above 14 ng/mL (Fig. 4). The curves illustrate the performance of various PSAD ratios in these three different PSA ranges. As the number UROLOGY / jANUARY1994 / VoLuME43, NUMBER1 ioo o15,,~ 10o o,o,j 902 o 90- 0.2 80i 80- 70: 70- 60- 60- •~" 5o-: .~ 50- 0.15 032 ,,~,~__j 0.46,~ ( : ~ 0.4 4o. Ul ~" 40u~ ao2 30- 39 patients 20- 20- 10- 142 patients 10- 0 0 ,,,i, 0 , , ~i,,,,i,,,,i,,,,I,,=,i,,,,i'~,,, 10 20 |, , ,, i,,, 30 40 50 60 70 80 False Positive Fraction (%) 90 100 ROC curve of PSAD in cases with serum PSA levels between 4. I and 10 ng/mL (DRE- TRUS-). F I G U R E 2. r/ 100 0 28 80 0.25 0a "~ i "%_[" 0 7O A oas, , ~ / 60 > 50 ~40. 30 -: 89 patients 20. 10. 0 ~'''1''''1 10 .... 20 I .... I .... I''''1 .... I''''1~''''1 30 40 50 60 70 80 False Positive Fraction (%) '''I ' ' 90 100 F I G U R E 3. ROC curve of PSAD in cases with serum PSA levels above I0 ng/mL (DRE- TRUS-). of cancers detected in patients with serum PSA between 10.1 and 14 ng/mL was small (7 cancers out of 50 patients), a ROC curve for this PSA range could not be drawn. The sensitivity and specificity of PSAD varied according to different cut-off points and different PSA ranges. Tables II, III, and IV show the effect of changes in the PSAD cut-off point on the test's sensitivity and specificity for patients with different PSA ranges. UROLOGY / JANUARY1994 / VOLU~tE43, NUMBER 1 ,,,i,, 0 lO '1 . . . . 20 I''~'1 .... I''''1''''1''''1''''1'''' 30 40 50 60 70 80 False Positive Fraction (%) 90 100 F I G U R E 4. ROC curve of PSAD in cases with serum PSA levels above 14 ng/mL (DRE- TRUS-). DRE+ T R U S - 0.2 90 v J Among the 60 patients in this group with serum PSA above 4 ng/mL, twenty-seven cancers (45%) were diagnosed. Eighteen cases (66.7%) were positive for cancer exactly at the same site of suspicion by DRE; of these, 15 showed additional involvement with cancer by systematic biopsies in other areas, 7 of which showed cancer in the contralateral side. Systematic biopsies detected the 9 other cases (33.3%) with cancer involvement in areas other than those suspected by DRE. Cancer was shown to be bilateral in 6 of these 9 cases, Fourteen patients had serum PSA levels between 4,1 and 10 ng/mL and PSAD less than 0.15. All three cancers in this subgroup were detected by directed biopsies, while no additional patients were found to have cancer by performing the systematic biopsies (Table V). D R E - TRUS+ Among the 69 patients in this group with serum PSA above 4 ng/mL, twenty-nine cancers (42%) were diagnosed. Twenty-two of these twenty-nine cancers (75.9%) were detected by directed biopsies, and twelve of them showed additional cancer involvement by systematic biopsies in areas other than the suspicious lesion. Seven other cancers (24,1%) were detected by systematic biopsies only; two of these were just beside the area of sus~ picion, while the remaining five were far from the suspicious area and thus would definitely have been missed without the systematic biopsies (e.g., 47 Effect of changes in PSAD cut-off point in patients with serum PSA levels between 4.1 and I0 ng/mL (DRE- TRUS-) TABLE II. I Upper limit [ Patients Patients of PSAD testing testing cut-off positive negative point' (potentially saved biopsies) True positives (cancers detected) False negatives (cancers missed) Sensitivity Specificity (%) (%) Effect of changes in PSAD cut-off point in patients with serum PSA levels above 14 ng/mL (DRE- TRUS-) TABLE IV. Upper limit of PSAD 'cut-off point' Patients testing positive Patients testing negative (potentially saved biopsies) True positives (cancers detected) False negatives (cancers missed) Sensitivity Specificity (%) (%) 0.00 142 0 23 0 100.0 0.0 0.00 39 0 13 0 100.0 0.0 0.05 139 3 23 0 100.0 2.5 0.10 37 2 13 0 100.0 7.7 0.10 112 30 22 1' 95.7 24.4 0.15 34 5 12 I* 92.3 15.4 0.13 80 62 22 1 95.7 51.3 0.20 29 10 12 1 92.3 34.6 0.14 72 70 21 2** 91.3 57.1 0.25 24 15 12 1 92.3 53.8 0.15 65 77 21 2 91.3 63.0 0.30 20 19 11 2** 84.6 65.4 0.16 55 87 19 4 82.6 69.7 0:35 15 24 8 5 61.5 73.1 0.17 49 93 17 6 73.9 73.1 0.40 73.1 *This case had a serum PSA value of 15.7 ng/mL, prostatic volume calculated by 0,20 81.5 *This case had a serum PSA value of 4.4 ng/mL, prostatic volume calculated by TRUS was 52.4 cc, and thus the patient had a PSAD of 0.084. The Gleason histologic score of the tumor was 7. Four of the 6 cores were positive for cancer, which involved 100 percent of one of the cores. **The second case had a serum PSA value of 4.3 ng/mL, prostatic volume calculated by TRUS was 33 cc, and thus the patient had a PSAD of 0.13. The Gleason histologic score was 4. One of the six cores was positive, with cancer involving only 10 percent of that core. TRUS was 140 cc, and thus PSAD was 0.112. The Gleason histologic score of the tumor was 7. One of the 6 cores was positive, with cancer involving less than 5 percent of that core. **The second case had a serum PSA value of 62.1 ng/mL, prostatic volume'calculated by TRUS was 223 cc, and thus PSAD was 0.278. The Gleason histologic score was 7. Two of the six cores were positive, with cancer involving 30 percent of both cores. Yield of systematic biopsies in patients with cancer suspected on DRE and~or TRUS with serum PSA levels between 4.1 and 10 ng/mL TABLE V. Effect of changes in PSAD cut-off point in patients with serum PSA levels between 10. I and 14 ng/mL (DRE- TRUS-) TABLE III. Upper limit of PSAD 'cut-off point' 0,00 Patients testing positive 50 Patients testing negative (potentially saved biopsies) 0 True positives (cancers detected) False negatives (cancers missed) 0 0.10 48 2 0 0.15 42 8 O.20 33 0.25 23 0.30 Sensitivity Specificity (%) (%) 100.0 0.0 100.0 4.7 1" 85.7 16,3 17 2** 71.4 34.9 27 4 42.9 53.5 No. of No. of cancers patients detected No. of cancers detected by directed biopsies No. of cancers detected by systematic biopsies only DRE + TRUS - <0,15 _>0.15 14 21 3 9 3 5 DRETRUS + < 0.15 _>0.15 14 19 0 10 0 0 8 2 DRE + TRUS + < 0.15 >_0.15 17 37 9 29 9 28 0 1 Total <0.15 >_0.15 45 77 12 48 12 41 0 4 69.8 *This case had a serum PSA value of 12 ng/mL, prostatic volume calculated by TRUS was 90.5 cc, and thus PSAD was 0.133. The Gleason histologic score of the tumor was 6. Four of the 6 cores were positive for cancer, which involved more than 50 percent of all cores. **The second case had a serum PSA value of 11.25 ng/mL, prostatic volume calculated by TRUS was 75 cc, and thus PSAD was 0.15. The Gleason histologic score was 5. One of the six cores was positive, with cancer involving less than 5 percent of that core. /+8 PSAD in 1 case, the right base was the area of suspicion, but the left apex was the area involved with cancer). No cancer was detected ill 14 patients with serum PSA between 4.1 and 10 ng/mL and PSAD less than 0.15 (Table V). UROLOGY / JANUARY•994 / VOLUME43, NUMBER 1 DRE+ TRUS+ Among the 131 patients in this group with serum PSA above 4 ng/mL, one hundred ten cancers (84%) were diagnosed; of these, one hundred five (95.5%) were found by directed biopsies and sixty-three of them showed additional cancer involvement by systematic biopsies in other areas. Five other cases (4.5%) were diagnosed by systematic biopsies alone; 4 of these were positive for cancer just beside the area of suspicion, and 1 was positive for cancer far from the area of suspicion. Seventeen patients had serum PSA between 4.1 and 10 ng/mL and a PSAD less than 0.15. All nine cancers in this subgroup were detected by directed biopsies, while no additional patients were found to have cancer by performing the systematic biopsies (Table V). In the whole population, systematic biopsies were the only method of detecting carcinoma in 64 cases (29.5% of d e t e c t e d cancers). T h e y showed additional areas involved with cancer other than those revealed by directed biopsies in 90 cases (41.5% of detected cancers). In comparing cancers found by systematic biopsies with those detected by directed biopsies, there was no significant difference in serum PSA levels (p = 0.114). The PSAD of cancers diagnosed by directed biopsies had a mean (_+ S.D.) of 0.71 __. 0.97, while those detected by systematic biopsies had a mean of 0.43 _ 0.48, showing a statistically significant difference between both groups (p = 0.032). Cancers found by systematic biopsies had a mean Gleason histologic score of 5.6 _+ 1.9, while those diagnosed by directed biopsies had a mean of 6.5 __. 1.5, representing a statistically significant difference (p = 0.0003). The percentages of cancers with a Gleason histologic score > 7 were 36.7 percent in cases detected by systematic biopsies, and 57.7 percent in those found by directed biopsies. It should be noted, however, that only 21.9 percent of cancers diagnosed by systematic biopsies were palpable, as compared with 85 percent of those detected by directed biopsies. The mean Gleason histologic score for palpable cancers was 6.6 _ 1.5, while that of impalpable cancers was 5.6 + 1.7, with a statistically significant difference between both groups (p = 0.0001). Of the palpable cancers, 60 percent had a Gleason histologic score >_ 7, as compared with 34.8 percent in cases of impalpable cancers. Six transition zone cancers were detected (2.8%) by biopsies directed toward suspicious areas in the transition zone. Their Gleason histologic scores ranged from 3 to 9 with a mean of 5.5 _+ 2.2, their serum PSA levels ranged from 4.6 to 48 UROLOGY / JANUARY1994 / VOLUME43, NUMBER 1 ng/mk with a mean of 19.3 _+ 15.0 ng/mk, and their PSAD ratios ranged from 0.23 to 0.88 with a mean of 0.61 _+0.29. Prostatic intraepithelial neoplasia (PIN) without associated cancer was diagnosed in 46 patients (8.6% of biopsies), 8 of which were of high grade. Their serum PSA levels ranged from 0.3 to 26.5 ng/mL with a mean of 8.4 +--4.9, and their PSAD ratios ranged from 0.01 to 0.41 with a mean of 0.2 +_0.1. PIN was associated with cancer in 59 cases (27.2% of cancers detected). Their serum PSA levels ranged from 4.4 to 80.8 ng/mL with a mean of 16.3 _+ 15.3, and their PSAD ratios ranged from 0.1 to 3.0 with a mean of 0.5 +_0.5. Patients with PIN alone had significantly lower serum PSA levels and PSAD ratios than those with PIN associated with cancer (p = 0.001 and p = 0.0004, respectively), but similar to those with other benign pathologies (p -- 0.6 and p = 0.98, respectively). COMMENT Our study cannot be considered to represent the result of routine screening for prostatic carcinoma, since all patients were referred to our center by urologists. However, it might help us to develop guidelines that could be used clinically in the early diagnosis of prostatic cancer, and could possibly assist in avoiding a significant number of biopsies. Studies by Cooner et al., in which all patients underwent DRE, serum PSA determination, and TRUS, provided comprehensive data supporting the value of integrating serum PSA and DRE as primary measures in any early detection program, being supplemented by TRUS in selected cases.l,12 It should be emphasized, however, that Cooner et al. only performed biopsies of suspicious lesions on DRE and/or TRUS. Other studies suggest that more cancers would probably have been detected had systematic biopsies been performed on every patient with elevated serum PSA levels. 13 In 1992, Catalona 7 proposed that all patients with serum PSA above 4 ng/mL should be subjected to systematic biopsies. To test Cooner's algorithm for the early detection of prostatic cancer, 6 and to evaluate Catalona's suggestion, 7 we prospectively performed systematic biopsies in all patients with serum PSA above 4 ng/mL, together with directed biopsies of suspicious lesions. By comparing areas under the ROC curves for serum PSA and PSAD in all patients, it was found that PSAD was a superior diagnostic tool for the detection of prostatic carcinoma. In the DRE- TRUS- patients with serum PSA between 4.1 and 10 ng/mL, the ROC curve for 49 PSAD (Fig. 2) indicates a cut-off point of approximately 0.15 in this PSA range. This is equivalent to the cut-off point suggested by Cooner, 6 below which he does not r e c o m m e n d performing systematic biopsies. However, it should be noted that Cooner's choice of 0.15 was based on directed biopsies in groups of patients with either suspicious DRE and/or TRUS. Only then did he apply his hypothesis to those with no suspicious findings on DRE and TRUS. He could not evaluate the group of interest (DRE- TRUS-) directly, as no biopsies were performed on these patients in his earlier studies. 1,4,12 Therefore, in our investigation, we systematically performed systematic biopsies on every patient in this group to assess the real value of PSAD. The data in Table II show that if we had limited systematic biopsies in the DRETRUS- group to patients with a PSAD > 0.15, we would have missed two cancers. This would have resulted in a detection rate of 14.8 percent instead of 16.2 percent, but only 3.1 patients would have required biopsies to find 1 with cancer as compared with 6.2. This would have spared biopsies in many patients (77 of 142 in our study) with a negligible decrease in the cancer detection rate. In patients with cancer suspected on DRE and/or TRUS with serum PSA levels between 4.1 and 10 ng/mL and with a PSAD below 0.15, all carcinomas were detected by directed biopsies (12 cancers of 45 patients) (Table V). Although systematic biopsies did not detect any additional patients with cancer, they showed bilateral involvement of the prostate in 4 of the 12 patients. This additional information might be important in patients who would be candidates for subsequent nervesparing prostatectomy. The data in Table V show that TRUS did not reveal any cancer in 14 patients with a serum PSA between 4.1 and 10 ng/mL, in the absence of positive findings on the DRE and with a PSAD < 0.15. This suggests that hypoechoic lesions have a very low specificity and are of questionable value under these conditions. Essentially, all authors agree that patients with serum PSA levels above 10 ng/mL require systematic biopsies. In DRE- TRUS- patients with serum PSA levels above 10 ng/mL, 4.5 required biopsies to find one cancer for a detection rate of 22.5 percent (Table I). In the other three groups, the detection rate was higher, and less patients required biopsies to find one cancer. We tried to assess the value of PSAD in D R E - TRUS- patients w i t h serum PSA levels above 10 ng/mL, to see if the n u m b e r of patients requiring biopsies could be safely decreased without jeopardizing cancer de50 tection. By drawing a ROC curve for PSAD in this group (Fig. 3), the best cut-off point was 0.3, but if applied to all patients in the group, six of twenty cancers would have been missed, sparing biopsies in 53 of 89 patients. In this case, 2.6 cases would have required systematic biopsies to find one cancer instead of 4.5, but the detection rate would drop to 15.7 percent instead of 22.5 percent. If we exclude patients with serum PSA between 10.1 and 14 ng/mL, and draw another ROC curve for PSAD in those with serum PSA levels above 14 ng/mL (Fig. 4), the best cut-off point is also 0.3. In this situation, two of thirteen cancers would have been missed, sparing biopsies in 19 of 39 patients, reducing the number of patients requiring biopsies to find one cancer from 3 to 1.8, and decreasing the detection rate from 33.3 percent to 28.2 percent (Table IV). As the number of cancers diagnosed in patients w i t h s e r u m PSA levels b e t w e e n 10.1 and 14 ng/mL was small (7 cancers of 50 patients), no proper ROC curve could be drawn for the PSAD ratios. Table III shows the n u m b e r of biopsies spared and cancers missed for the different PSAD cut-off points. It indicates that at a PSAD of 0.2, two cancers would have been missed and biopsies spared in 17 patients. As the number of cancers detected in this category was small, more cases are needed to conclude if there is any value of a PSAD cut-off point in this serum PSA range. Our results are based on a single set of systematic biopsies. It is probable that we would detect more cancers if systematic biopsies were repeated 7 in patients with serum PSA levels above 10 ng/mL who had a non-malignant result from the first set of biopsies. This evaluation is presently underway. Systematic biopsies were the only method of detecting cancer in 29.5 percent of the malignant cases. They added further information to the diagnosis in 41.5 percent of cases in which cancer was found with directed biopsies, by showing multifocal lesions. Regarding the clinical significance of cancers detected by systematic biopsies, there appeared to be a significantly lower Gleason histologic score in this group as compared with cancers diagnosed by directed biopsies (p = 0.0003). However, this may be due to the fact that most of the cancers detected by directed biopsies were digitally palpable, while most of those found by systematic biopsies were impalpable. On comparing palpable with impalpable cancers, the Gleason histologic score was significantly higher for the palpable cancers (p = 0.0001). On comparing Gleason histologic UROLOGY / JaNuaeY 1994 / VOLUME43, NUMBER1 scores of i m p a l p a b l e cancers only, there w a s n o statistically s i g n i f i c a n t difference b e t w e e n t h o s e d e t e c t e d b y s y s t e m a t i c biopsies a n d those f o u n d b y directed biopsies (p = 0.85). This m a y be related to the fact that palpable cancers have a larger v o l u m e t h a n i m p a l p a b l e c a n c e r s ) H o w e v e r , this issue c o u l d n o t be a d d r e s s e d in o u r s t u d y as the patients w e r e referred to o u r center b y several different urologists, a n d the data are n o t c o m p l e t e reg a r d i n g p a t i e n t m a n a g e m e n t a n d t u m o r size in radical p r o s t a t e c t o m y s p e c i m e n s (if done). W e d i d n o t u s e s y s t e m a t i c b i o p s i e s in an att e m p t to detect transition z o n e cancers. 5 T h e six (2.8%) transition z o n e cancers d i a g n o s e d in o u r s t u d y w e r e all f o u n d b y d i r e c t e d biopsies. T h i s low yield of transition z o n e t u m o r s is similar to that r e p o r t e d b y o t h e r investigators.1 CON CLUSION As a result of o u r study, we c o n c l u d e that PSAD is a useful diagnostic tool for the early detection of prostatic c a r c i n o m a . In the D R E - T R U S - g r o u p of p a t i e n t s w i t h a PSA r a n g e b e t w e e n 4.1 a n d 10 ng/mL, o u r data s u p p o r t the n o t i o n that s y s t e m atic biopsies can be safely limited to cases w i t h a PSAD > 0.15. Patients n o t s u b j e c t e d to s y s t e m a t i c biopsies s h o u l d be followed to search for a n y potentially u n d i a g n o s e d cancers. In patients in the s a m e PSA r a n g e w i t h c a n c e r s u s p e c t e d o n DRE a n d / o r TRUS a n d w i t h a PSAD < 0.15, s y s t e m a t i c biopsies did not detect any additional patients w i t h cancer, however, they are useful in d e t e r m i n ing if c a r c i n o m a exists o n the o p p o s i t e side. O u r data s u g g e s t t h a t in p a t i e n t s w i t h s e r u m PSA levels a b o v e 14 n g / m L a n d w i t h o u t s u s p e c t e d cancer o n b o t h DRE a n d TRUS, s y s t e m a t i c b i o p sies c o u l d b e safely limited to those w i t h a PSAD o f _> 0.3. H o w e v e r , c l i n i c a l j u d g m e n t m a y b e n e e d e d in i n d i v i d u a l cases. A larger s t u d y is req u i r e d before this a p p r o a c h c o u l d b e c o n s i d e r e d for a d o p t i o n . In p a t i e n t s w i t h s e r u m PSA levels b e t w e e n 10.1 a n d 14 n g / m L , the role of PSAD c o u l d n o t be evaluated due to the small n u m b e r of cancers detected in this category in o u r investigation. Michel Bazinet, M.D. Department of Urology The Montreal General Hospital 1650 Cedar Avenue Montreal Quebec H3G 1A4 Canada REFERENCES 1. Cooner WH, Mosley BR, Rutherford CL Jr, Beard JH, Pond HS, Terry WJ, Igel TC, and Kidd DD: Prostate cancer UROLOGY / JANUARY•994 / VOLUMe43, NUM13~R1 detection in a clinical urological practice by ultrasonography, digital rectal examination and prostate specific antigen. J Urol 143: 1146-1154, 1990. 2. Catalona WJ, Smith DS, Ratliff TL, Dodds KM, Coplen DE, Yuan JJJ, Petros JA, and Andriole GL: Measurement of prostate-specific antigen in serum as a screening test for prostate cancer. N EnglJ Med 324: 1156-1161, 1991. 3. Stamey TA, Yang N, Hay AR, McNealJE, Freiha FS, and Redwine E: Prostate-specific antigen as a serum marker for adenocarcinoma of the prostate. N Engl J Med 317: 909-916, 1987. 4. Benson MC, Whang IS, Olsson CA, McMahon DJ, and Cooner WH: The use of prostate specific antigen density to enhance the predictive value of intermediate levels of serum prostate specific antigen. J Urol 147: 817-821, 1992. 5. Hodge KK, McNeal JE, Terris MK, and Stamey TA: Random systematic versus directed ultrasound guided transrectal core biopsies of the prostate. J Urol 142: 71-75, 1989. 6. Cooner WH, Babaian RJ, and Bostwick DG: Prostate cancer detection using DRE, prostate specific antigen and prostate-ultrasonography. Postgraduate Course #77, AUA Annual Meeting (24A), Washington, DC, May 14, 1992. 7. Catalona WJ: Prostate specific antigen as a screening test for prostate cancer. State of the Art Lecture. AUA Annual Meeting (63A), Washington, DC, May 10, 1992. 8. Lee F, Gray JM, McLeary RD, Meadows TR, Kumasaka GH, Borlaza GS, Straub W, Lee F Jr, Solomon MH, McHugh TA, and Wolf RM: Transrectal ultrasound in the diagnosis of prostate cancer: location, echogenicity, histopathology, and staging. Prostate 7: 117-129, 1985. 9. Evaluating diagnostic procedures, in Dawson-Saunders B, and Trapp RG (Eds): Basic and Clinical Biostatistics, East Norwalk, Appleton and Lange, 1990, chapt 13, pp 229-244. 10. Hanley JA, and McNeil BJ: The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 143: 29-36, 1982. 11. Hanley JA, and McNeil BJ: A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 148: 839-843, 1983. 12. Cooner WH: Prostate specific antigen, digital rectal examination, and transrectal ultrasonic examination of the prostate in prostate cancer detection. Monogr Urol 12: 3-13, 1991. 13. Vallancien G, Prapotnich D, Veillon B, Brisset JM, and Andre-BougaranJ: Systematic prostatic biopsies in 100 men with no suspicion of cancer on digital rectal examination. J Urol 146: 1308-1312, 1991. EDITORIAL COMMENT The diagnosis of prostatic carcinoma rests entirely on microscopic tissue analysis. Suspicion that cancer may exist comes from abnormal physical findings, elevated serum markers, or abnormalities seen with imaging modalities. All of these suffer from low specificity. Suspicious digital rectal examination (DRE) findings almost always lead to biopsy. The advent of serum prostate-specific antigen (PSA) and transrectal ultrasonic examination of the prostate (TRUS) introduces the question as to which patients need biopsy should either of these be abnormal. Even if all available diagnostic tests were used, it would be impossible to detect every prostatic carcinoma, including those that would be destined to become clinically significant. Use of TRUS without DRE or PSA abnormality adds little if biopsy is directed only at visually suspicious areas. Because of the high prevalence of prostatic carcinoma in men over age 51