Journal of Computer Assisted Tomography 26(3):432–437 © 2002 Lippincott Williams & Wilkins, Inc., Philadelphia
Likelihood of Prostate Cancer Based on Prostate-Specific Antigen Density by MRI: Retrospective Analysis Ullrich G. Mueller-Lisse, Ulrike L. Mueller-Lisse, Steffen Haller, Peter Schneede, Juergen E. Scheidler, Nicolaus Schmeller, Alfons G. Hofstetter, and Maximilian F. Reiser
Objective: As a screening test for prostate cancer (PCA), prostate-specific antigen (PSA) may induce unnecessary prostate biopsy in patients with PSA 4.1–10.0 ng/ml. PCA detection may be delayed in patients with PSA ⱕ4.0 ng/ml. MRI-based PSA density of the prostate (PSAD) and of the prostatic transitional zone (PSAT) could improve differentiation of PCA and benign prostatic hyperplasia. Material and Methods: Total prostate and transitional zone volumes were planimetrically determined in axial, T2-weighted fast spin echo MR images of the prostate. Serum PSA concentration was measured with an automated standardized microparticle enzyme immune assay. PSAD and PSAT were calculated in 17 patients with clinically significant PCA and 42 patients with benign prostatic hypertrophy (BPH) (66 ± 6 versus 64 ± 8 years, p ⳱ 0.2410, t test) who had PSA levels ⱕ10.0 ng/ml. Results: For differentiation of BPH and PCA, PSA alone above the optimal cutoff level of 4.2 ng/ml showed an odds ratio for PCA of 6.7 (95% confidence interval [CI], 1.9–23.2). PSAD showed an odds ratio for PCA of 71.3 (95% CI, 11.8–430.9) above the optimal cutoff level of 0.07 ng/ml/cc. PSAT demonstrated an odds ratio for PCA of 320.0 (95% CI, 27.1–3781.4) above the optimal cutoff level of 0.15 ng/ml/cc. Conclusions: In patients with PSA ⱕ10.0 ng/ml, MRI-based PSAD and PSAT appear to improve differentiation of prostate cancer and BPH and are feasible to reduce the frequency of unnecessary prostate biopsy. Index Terms: Prostate—Magnetic resonance imaging—Prostate-specific antigen—Prostate—Transitional zone—Prostate-specific antigen density.
nation (TRUS) (2,3). Magnetic resonance imaging (MRI)-based prostate volume estimates require a single, axial, T2-weighted sequence (4,5). They demonstrate a higher correlation with prostate specimen wet weights than TRUS-based volume estimates (4), and an intra- and interobserver reproducibility of ±5% (4,5). MRI-based calculation of PSAD and PSAT could decrease the proportion of patients with mere BPH who would undergo unnecessary biopsy if patients with PCA were recognized. We retrospectively determined prostate volumes by planimetry in T2-weighted MR images of the prostate and compared the differentiation of BPH and PCA by PSA alone, PSAD, and PSAT. In a case-control study, we evaluated MR images obtained with the body coil or body phased array coil in a limited selection of patients with clinical PCA (cases) or BPH (controls) who had PSA levels ⱕ10.0 ng/ml.
INTRODUCTION The serum concentration of prostate-specific antigen (PSA) represents a valuable tumor marker. Biopsy has been recommended for all patients with PSA >4.0 ng/ml to rule out prostate cancer (PCA). However, PCA is found in a proportion of patients with PSA <4.0 ng/ml, and benign prostatic hyperplasia (BPH) alone accounts for 50–85% of cases with PSA between 4.1 ng/ml and 10.0 ng/ml (1). The utility of PSA density of the entire prostate (PSAD) and PSA density of the transitional zone (PSAT), the ratios of PSA per volume unit of the respective prostatic compartments, is controversial when volume estimates are based on transrectal ultrasound examiFrom the Departments of Radiology (U. G. Mueller-Lisse, S. Haller, J. E. Scheidler, and M. F. Reiser) and Urology (U. L. Mueller-Lisse, P. Schneede, N. Schmeller, and A. G. Hofstetter), Klinikum Grosshadern, University of Munich, Munich, Germany. Address correspondence to Dr. U. G. Mueller-Lisse, Department of Radiology, Klinikum Grosshadern, University of Munich, 15, Marchionini St., D-81377 Muenchen, Germany. E-mail: [email protected]
MATERIAL AND METHODS Axial, T2-weighted fast spin echo (FSE) MR images of the prostate were reviewed in all patients with PSA 432
LIKELIHOOD OF PROSTATE CANCER BASED ON PSA DENSITY BY MRI ⱕ10.0 ng/ml (Abbott IMx MEIA; Abbott Laboratories, N. Chicago, IL, U.S.A.) (minimum detection level 0.5 ng/ml) who had undergone MRI of the prostate with a body coil or body phased array coil system at 1.0 T (“Magnetom Impact,” Siemens Medical Systems, Erlangen, Germany) (36 patients ) or 1.5 T (“Magnetom Vision,” Siemens Medical Systems) (23 patients) between March 1994 and August 1997. The rationale for a retrospective evaluation of body coil and body phased array coil examinations instead of endorectal coil examinations was that application of the former is less invasive and less costly. Therefore, body coil or body phased array coil MRI would appear more suitable to determine PSA density of the prostate in a clinical trial or clinical setting. In all, 59 consecutive patients met inclusion criteria for the study. All PSA tests were performed within 1 month prior to MRI. Prostate biopsy had been performed at least 1 week prior to MRI and was negative in benign prostatic hyperplasia (BPH) patients with PSA >4.0 ng/ml. Patients had no previous operative procedure, radiation, or hormone therapy of the prostate, or urinary catheterization at the time of PSA measurement, MRI, or prostate biopsy, and had no acute prostatitis within the previous month. In anticipation of unclear accuracy and interobserver reproducibility, results of digital rectal examination were not considered. Seventeen patients (average age, 66 ± 6 years, PSA ⱕ4.0 ng/ml, n ⳱ 6, 4.1–10.0 ng/ml, n ⳱ 11) had undergone MRI as part of their clinical work-up for biopsyproved PCA. Ten patients had radical prostatectomy (pT2, n ⳱ 9; pT3, n ⳱ 1; all pN0 M0), 4 had transurethral resection of the prostate (T1b, n ⳱ 4), and 3 had previously unrecognized positive nodes at lymphadenectomy, such that prostatectomy was abandoned. Forty-two patients (average age, 64 ± 8 years, not significantly different from PCA patients, p ⳱ 0.2410, Student t test; PSA ⱕ4.0 ng/ml, n ⳱ 32, 4.1–10.0 ng/ml, n ⳱ 10) had BPH with a nonsuspicious clinical and PSA history, and without evidence of PCA at clinical examination and biopsy. These patients had undergone MRI to determine size and shape of the prostate as part of their clinical
work-up because they were seeking alternative, nonresective treatments to transurethral resection for symptomatic BPH. Subsequently, three patients underwent transurethral resection of the prostate or suprapubic adenomectomy without evidence of PCA. Thirty-four patients who underwent laser coagulation therapy of BPH and 5 patients who did not undergo surgical treatment were followed-up clinically, including repetitive clinical examination, PSA testing, and MRI of the prostate over 6–12 months, without evidence of PCA. Laser coagulation therapy of BPH was an alternative treatment offered by our hospital on a clinical trial basis. The local hospital ethics committee approved the clinical trial. Axial, T2-weighted FSE MR images were obtained with TR 4,700 ms, TE 112 ms, field-of-view (FOV) 250 × 250 mm, matrix 180 × 256 pixels, slice thickness 4 mm, interslice gap 1 mm, stack of 13 contiguous slices, data acquisition time 4 min at 1.0 T. Parameters at 1.5 T were TR 4,700 ms or 5,300 ms, TE 99 ms or 112 ms, FOV 135 × 135–250 × 250 mm, matrix 240 × 256 or 256 × 512 pixels, slice thickness 3 mm or 4 mm, interslice gap 0 or 1 mm, stack of 13–23 contiguous slices, data acquisition time 2–6 min. In each case, the entire prostate was covered from apex to base. Total prostate volume and transitional zone volume were determined by planimetry, through tracing of the prostatic capsule or pseudocapsule in the MR images (Fig. 1), respectively (4,5). Software and hardware components commercially available with purchase of the original MRI systems were used for evaluation. MR images were displayed on the screen of the MR console, at the largest image display setting offered by the display menu, but without additional magnification. Tracing was performed with the mouse. The “irregular region of interest” evaluation mode was used as provided by the manufacturer as part of the standard MRI system software. Additional planimetry devices were not used. Reproducibility of planimetry results was tested in 20 cases (BPH, n ⳱ 12; PCA, n ⳱ 8) both between two different observers and for the same observer. The same observer performed all tracings that entered the evaluation of PSA density. Division of the PSA level by the respective pros-
FIG. 1. Tracing of prostatic compartments in a 71-year-old patient with symptomatic prostatic hyperplasia. At 8.9 ng/ml, PSA was in the “gray zone” (4.1–10.0 ng/ml), but prostate volume was large. Pretherapeutic prostate biopsy revealed moderately differentiated cancer. Prostatectomy was abandoned because of lymphatic micrometastasis. A: Axial, T2-weighted MR image at prostatic midgland level. B: Tracing of the prostatic capsule for planimetric determination of total prostate volume (89.0 cc). Prostate-specific antigen (PSA) density of the prostate was 0.10 ng/ml/cc, suggesting prostate cancer. C: Tracing of the prostatic pseudocapsule for planimetric determination of transitional zone volume (60.7 cc). PSA density of the transitional zone was 0.15 ng/ml/cc, being at the borderline for prostate cancer.
J Comput Assist Tomogr, Vol. 26, No. 3, 2002
U. G. MUELLER-LISSE ET AL.
tatic volume calculated PSA density of the prostate (PSAD) and PSA density of the transitional zone (PSAT). In anticipation of possible confounding by case selection criteria and nonrepresentative frequency of prostate cancers in this retrospective study, differentiation of BPH and PCA by PSA alone, PSAD, and PSAT was expressed as odds ratios rather than sensitivity and specificity (see Appendix). For statistical analysis of dichotomized data in 2 × 2 tables, the 2 test with Yates’s continuity correction was applied at a significance level of p < 0.05. RESULTS Delineation of both the prostatic capsule and pseudocapsule was sufficient in all cases. Reproducibility of total prostate volume was ±3.8% (0.3–6.9%) between observers and ±2.5% (0.7–8.8%) for the same observer. Reproducibility of transitional zone volume was ±5.4% (3.3–13.2%) between observers and ±3.3% (0.6–11.5%) for the same observer. At the cutoff level of 4.0 ng/ml that is frequently used to select patients at risk for PCA, PSA recognized 11/17 PCAs (65%) and 32/42 BPHs (76%). In this study, the optimal cutoff level for PSA was 4.2 ng/ml. At this cutoff level, PSA recognized 11/17 PCAs (65%) and 33/42
BPHs (79%); 2 for the differentiation of PCA and BPH was 8.2760, p ⳱ 0.004. The odds ratio for PCA at PSA levels exceeding 4.2 ng/ml was 6.7 (95% confidence interval [CI], 1.9–23.2). At the optimal cutoff level for PSAD of 0.07 ng/ml/cc, 15/17 PCAs (88%) and 38/42 BPHs (90%) were recognized, including four of six patients with PCA and PSA ⱕ4.0 ng/ml; 2 for the differentiation of PCA and BPH was 30.8304, p < 0.001. The odds ratio for PCA at PSAD exceeding 0.07 ng/ml/cc was 71.3, (95% CI, 11.8– 430.9). However, the performance of PSAD was not significantly better than PSA at the optimum cutoff level (53/59 vs. 44/59 correct results, 2 3.7074, p ⳱ 0.055). At the optimal cutoff level for PSAT of 0.15 ng/ml/cc, 16/17 PCAs (94%) and 40/42 BPHs (95%; 2 for the differentiation of PCA and BPH was 41.4590, p < 0.001) were recognized, including all patients with PCA and PSA ⱕ4.0 ng/ml. The odds ratio for PCA at PSAT exceeding 0.15 ng/ml/cc was 320.0 (95% CI, 27.1–3781.4). The only PCA patient missed had a borderline PSAT of 0.15 ng/ml/cc (Figs. 1 and 2), but a positive PSAD of 0.10 ng/ml/cc, associated with micrometastasis of PCA in a single lymph node. The number of correct calls by PSAT was slightly, but not significantly better than by PSAD (56/59 vs. 53/59 correct results, 2 0.4811, p ⳱ 0.48). However, the performance of PSAT was significantly better than PSA at the optimum cutoff level (56/59 vs. 44/59 correct results, 2 7.9322, p < 0.005).
FIG. 2. Differentiation of patients with benign prostatic hyperplasia (×) and patients with prostate cancer (䊊) by prostate-specific antigen (PSA), PSA density of the prostate (PSAD), and PSA density of the transitional zone of the prostate (PSAT). Short, transverse lines mark respective cutoff levels. Logarithmic scale for the y axis was used to accommodate all data.
J Comput Assist Tomogr, Vol. 26, No. 3, 2002
LIKELIHOOD OF PROSTATE CANCER BASED ON PSA DENSITY BY MRI DISCUSSION The introduction of PSA as a screening tool, with an optimum cutoff level of 4.0 ng/ml, has improved the early detection of PCA, which is now newly diagnosed in approximately 200,000 men per year in the United States (1). However, to rule out PCA in the gray zone of PSA between 4.1 and 10.0 ng/ml, a great number of men with BPH would undergo unnecessary prostate biopsy. This would be associated with cost, discomfort (minor pain, 92–95%), risks of bleeding (macrohematuria, 11–47%, hematospermia, 45–89%), infection (urinary tract infection, 0.8–8%, fever, 4%, including hospitalization in 0.6– 2%), and tumor cell spread (0.1–0.2%) (6–13). On the other hand, the discrepancy between a negative PSA test and a suspicious rectal examination may prevent early PCA detection in some men (1). While calculation of PSAD or PSAT has increased specificity in the differentiation of prostate cancer and BPH in some studies when based on TRUS (2,14–16), other authors found no advantage over PSA alone (3,17). Despite controversial TRUS results, we found both MRIbased PSAD and PSAT superior to PSA alone in the differentiation of PCA and BPH in our limited selection of patients with PSA ⱕ10.0 ng/ml. Our retrospective case-control study of 59 patients confirms observations made in the initial, retrospective, TRUS-based study by Zlotta et al. (2) in 88 patients with PCA and 74 patients with BPH with PSA levels ⱕ10 ng/ml. At optimum cutoff levels for the differentiation of BPH from PCA, Zlotta et al. (2) report a sensitivity of 74.3% and specificity of 65.9% for PSAD (cutoff level 0.15 ng/ml/cc), and a sensitivity of 93.2% and specificity of 89.7% for PSAT (cutoff level 0.35 ng/ml/cc). The severe differences between the optimal cutoff values for TRUS-based (2) and MRI-based PSAD measurements are unlikely to be due only to different definitions of the respective underlying volumes. We referred to that part of the prostate that is outlined by a low signal intensity pseudocapsule on axial, T2-weighted MR images (4,5) as the transitional zone, although it most likely comprises tissue of the transitional zone, central zone, and periurethral tissue. It is unclear if the transitional zone at TRUS, defined by Zlotta et al. (2) as being outlined by a distinct layer of fibrous tissue, actually represents the same entity, or a less comprehensive volume. This could be one reason for the severe difference between the PSAD cutoff values for the central prostate volumes between the two studies. However, the confines of the entire prostate gland can be expected to be the same for both TRUS and MRI. Nonetheless, the differences between cutoff levels for TRUS-based (2) and MRI-based PSAD of the entire prostate are of the same order as those for PSAT of the central volumes. Therefore, potential differences in volume definition alone cannot explain the differences between the TRUS study (2) and our observations with MRI. Since TRUS-based prostate volumes in the study by Zlotta et al. (2) were estimated nonplanimetrically, with
the ellipsoid formula, the actual method and reliability of volume estimation may explain different optimum cutoff levels. This hypothesis is supported by another TRUSbased study (15) that determined an optimal cutoff value for PSAT of 0.17 ng/ml/cc, more similar to the optimum cutoff we found for MRI-based PSAT. The reliability and reproducibility of prostate volume assessment by TRUS depends not only on the volumetry method applied (18,19), but also on the individual experience of the examiner (20). Differences between two consecutive prostate volume estimates range from 15.5–25.5% for nonplanimetric measurements, and from 1.7–11.4% for planimetric measurements (18,19,21). Besides its potential advantage of independence from the examiner, MRI has demonstrated high intra- and interobserver reproducibility of planimetric volumetry of the prostate, with deviations ⱕ5% in the literature (4,5) and in this study. MRI has shown superior correlation of prostate volumes assessments with actual specimen wet weights when compared with TRUS (4). Another factor that may have contributed to the differences between optimum cutoff levels determined for MRI-based PSAD in this study and previous TRUSbased PSAD results (2) may be found in the underlying PSA test. We only included patients whose PSA measurement had been performed in our hospital, with one particular automated monoclonal microparticle enzyme immune assay (MEIA). However, in patients with PSA concentrations exceeding 1.0 ng/ml, MEIA PSA levels have been found to be approximately 1.73 times lower than PSA levels determined by radioimmunometric assay (ELSA) when carried out in the same patients, despite high correlation (r ⳱ 0.93) between results of both tests (22). Compared with several studies involving TRUS for the calculation of PSAD or PSAT (2,3,14–17), the sample size in our study was relatively small. Inference to larger populations may be somewhat limited by the fact that patients were selected retrospectively from two different populations, including patients who had MRI of the prostate for staging of prostate cancer and patients who had MRI of the prostate in preparation of alternative treatments of benign prostatic hyperplasia. We limited the selection to patients with a PSA level of up to 10.0 ng/ml, to account for both the “gray zone” of PSA between 4.0 and 10.0 ng/ml, and for those patients with prostate cancer who present with PSA levels lower than 4.0 ng/ml. Also, we restricted the analysis to patients with body coil MRI or body phased array coil MRI of the prostate, because of lesser invasiveness, shorter times of preparation preceding MRI, and lesser cost of examination. Nonetheless, the optimum cutoff level for PSA of 4.2 ng/ml and the average ages of 66 and 64 years in patients with PCA and BPH, respectively, imply that key characteristics of our study populations were reasonably close to the general population affected by PCA and BPH. Also, our results are in keeping with those of previous studies (2,14–16) and new studies in larger patient populations that report a moderate advantage of PSAD J Comput Assist Tomogr, Vol. 26, No. 3, 2002
U. G. MUELLER-LISSE ET AL.
and PSAT over PSA alone in the discrimination of BPH and PCA (23,24), even when PSA cutoff is corrected for patient age (25). Whether PSAD or PSAT is more promising for the differentiation of BPH and PCA than the relative concentration of free PSA remains the subject of discussion (23,24,26). The axial, T2-weighted MR images underlying our MRI-based calculations of PSAD and PSAT were obtained within 2–6 minutes. Except for localizer images prior to the sequence, with an acquisition time of less than 1 minute, no other preparation was necessary. MRI measurements for PSAD determination were completely noninvasive, and did not require the use of an endorectal probe, anesthesia, or contrast media. In contrast to TRUS, a trained technologist could perform both the MRI scan and the planimetry. In view of the short preparation times and MRI scan times necessary for PSAD and PSAT, the limited scope of resources involved, and the higher accuracy of MRI (4), the view that TRUS is preferable to MRI for prostate volumetry because of its lower cost (27) probably cannot be upheld. CONCLUSION In conclusion, MRI-based PSAD and PSAT are more promising than PSA alone in the prevention of unnecessary prostate biopsy in patients with PSA ⱕ10.0 ng/ml. PSAD and PSAT require a single, axial, T2-weighted prostate scan, which takes 2–6 minutes of acquisition time on regular MR systems. Furthermore, PSAD and PSAT appear feasible as an adjunct to a full MRI examination of the prostate. This would be useful to assess the presence of prostate cancer in patients with PSA levels in the gray zone who either seek minimally invasive or noninvasive therapy for symptomatic BPH, or who present with a suspicion of cancer and a negative prostate biopsy result. The retrospective nature and limited scope of our study prevent detailed analysis of diagnostic differences between MRI-based PSAD and PSAT. A prospective study of larger scale would be desirable to characterize patients in whom MRI-based PSAD and PSAT can prevent or postpone potentially unnecessary prostate biopsy. APPENDIX Use of the Odds Ratio in a Case-Control Study The principal data underlying a case-control study such as this one can be summarized in 2 × 2 tables as below (Table 1). TABLE 1. 2 × 2 table sample
Risk factor present Risk factor absent
Controls (no disease)
J Comput Assist Tomogr, Vol. 26, No. 3, 2002
Risk factors in this study are PSA, PSAD, and PSAT; the disease is prostate cancer as opposed to BPH (no disease). If the 2 × 2 table represented data obtained from a cohort study, the incidence of the disease in those with the risk factor would be a/(a + b), and the relative risk would be [a/(a + b)]/[c/(c + d)]. The sensitivity of the risk factor for the disease would be a/(a + c), and its specificity would be d/(d + b). However, it is not appropriate to compute these data in this way because the two samples (i.e., cases and controls) are not drawn from the population in the same proportions. Usually, there are many fewer cases than controls in the population, while there are roughly similar numbers of cases and controls in the study. Instead, relative risk in a case-control study can be approximated by the odds ratio. The odds ratio is defined as the ratio of the odds of disease in those with the risk factor, a/b, to the odds of the disease in those without the risk factor, c/d. This can be arranged as the cross product of the 2 × 2 table, (a*d)/(c*b). The odds ratio in the study is a close approximation of the relative risk in the population if the following conditions are fulfilled. The cases have to be representative of the population (i.e., they have the same prevalence of the risk factor as patients with the disease in the population). The controls have to be similarly representative (i.e., they have the same prevalence of the risk factor as patients without the disease in the population). The prevalence of the disease (in this study, clinically significant prostate cancer) in the population has to be low and the sampling error (systematic and random) has to be small. [Modified from Appendix 8.B., page 206, in: Hulley SB, Cummings SR. Designing Clinical Research. An Epidemiological Approach. Baltimore: Williams & Wilkins, 1988.] REFERENCES 1. Catalona WJ, Richie JP, deKernion JB, et al. Comparison of prostate specific antigen concentration versus prostate specific antigen density in the early detection of prostate cancer: receiver operating characteristic curves. J Urol 1994;152:2031–6. 2. Zlotta AR, Djavan B, Marberger M, et al. Prostate specific antigen density of the transition zone: a new effective parameter for prostate cancer prediction. J Urol 1997;157:1315–21. 3. Lin DW, Gold MH, Ransom S, et al. Transition zone prostate specific antigen density: lack of use in prediction of prostatic carcinoma. J Urol 1998;160:77–81. 4. Rahmouni A, Yang A, Tempany CM, et al. Accuracy of in-vivo assessment of prostatic volume by MRI and transrectal ultrasonography. J Comput Assist Tomogr 1992;16:935–40. 5. Mueller-Lisse UG, Heuck AF, Schneede P, et al. Postoperative MRI in patients undergoing interstitial laser coagulation thermotherapy of benign prostatic hyperplasia. J Comput Assist Tomogr 1996;20:273–8. 6. Aus G, Hermansson CG, Hugosson J, et al. Transrectal ultrasound examination of the prostate: complications and acceptance by patients. Br J Radiol 1993;71:457–9. 7. Castineiras J, Varo C, Castro C, et al. Complications of ultrasoundguided transperineal puncture biopsy of the prostate. Actas Urol Esp 1995;19:544–8. 8. Deliveliotis C, John V, Louoras G, et al. Multiple transrectal ultrasound guided prostatic biopsies: morbidity and tolerance. Int Urol Nephrol 1999;31:681–6.
LIKELIHOOD OF PROSTATE CANCER BASED ON PSA DENSITY BY MRI 9. Desmond PM, Clark J, Thompson IM, et al. Morbidity with contemporary prostate biopsy. J Urol 1993;150:1425–6. 10. Gustafsson O, Norming U, Nyman CR, et al. Complications following combined transrectal aspiration and cor biopsy of the prostate. Scand J Urol Nephrol 1990;24:249–51. 11. Naughton CK, Ornstein DK, Smith DS, et al. Pain and morbidity of transrectal ultrasound guided prostate biopsy: a prospective randomized trial of 6 versus 12 cores. J Urol 2000;163:168–71. 12. Rietbergen JB, Kruger AE, Kranse R, et al. Complications of transrectal ultrasound-guided systematic sextant biopsies of the prostate: evaluation of complication rates and risk factors within a population based screening program. Urology 1997;49:875–80. 13. Rodriguez LV, Terris MK. Risks and complications of transrectal ultrasound guided prostate needle biopsy: a prospective study and review of the literature. J Urol 1998;160:2115–20. 14. Horninger W, Reissigl A, Klocker H, et al. Improvement of specificity in PSA-based screening by using PSA-transition zone density and percent free PSA in addition to total PSA levels. Prostate 1998;37:133–7. 15. Kurita Y, Terada H, Masuda H, et al. Prostate specific antigen (PSA) value adjusted for transition zone volume and free PSA (gamma-seminoprotein)/PSA ratio in the diagnosis of prostate cancer in patients with intermediate PSA levels. Br J Urol 1998;82: 224–30. 16. Maeda H, Arai Y, Ishitoya S, et al. Prostate specific antigen adjusted for the transition zone volume as an indicator of prostate cancer. J Urol 1997;158:2193–6. 17. Gohji K, Nomi M, Egawa S, et al. Detection of prostate carcinoma using prostate specific antigen, its density, and the density of the transition zone in Japanese men with intermediate serum prostate specific antigen concentrations. Cancer 1997;79:1969–76. 18. Tong S, Cardinal HN, McLoughlin RF, et al. Intra- and interobserver variability and reliability of prostate volume measure-
ment via two-dimensional and three-dimensional ultrasound imaging. Ultrasound Med Biol 1998;24:673–81. Bazinet M, Karakiewicz PI, Aprikian AG, et al. Reassessment of nonplanimetric transrectal ultrasound prostate volume estimates. Urology 1996;47:857–62. Collins GN, Raab GM, Hehir M, et al. Reproducibility and observer variability of transrectal ultrasound measurements of prostatic volume. Ultrasound Med Biol 1995;21:1101–5. Aarnink RG, De La Rosette JJ, Debruyne FM, et al. Reproducibility of prostate volume measurements from transrectal ultrasonography by an automated and a manual technique. Br J Urol 1996;78:219–23. Sanchez de la Muela P, Adot JM, Povo J, et al. A comparative analysis of monoclonal assays for the determination of PSA: radioimmunometric assay (ELSA) vs. microparticle immunoenzyme assay (MEIA). Actas Urol Esp 1993;17:569–73. Djavan B, Zlotta AR, Remzi M, et al. Total and transition zone prostate volume and age: how do they affect the utility of PSAbased diagnostic parameters for early prostate cancer detection? Urology 1999;54:846–52. Egawa S, Suyama K, Takashima R, et al. Prospective evaluation of prostate cancer detection by prostate-specific antigen-related parameters. Int J Urol 1999;6:493–501. Meshref AW, Bazinet M, Trudel C, et al. Role of prostate-specific antigen density after applying age-specific prostate-specific antigen reference ranges. Urology 1995;45:972–79. Catalona WJ, Southwick PC, Slawin KM, et al. Comparison of percent free PSA, PSA density, and age-specific PSA cutoffs for prostate cancer detection and staging. Urology 2000;56:255–60. Tewari A, Indudhara R, Shinohara K, et al. Comparison of transrectal ultrasound prostatic volume estimation with magnetic resonance imaging volume estimation and surgical specimen weight in patients with benign prostatic hyperplasia. J Clin Ultrasound 1996; 24:169–74.
J Comput Assist Tomogr, Vol. 26, No. 3, 2002