| Japanese Journal of Clinical Oncology | Pages |
Determination of Reference Values for Total PSA, F/T and PSAD According to Prostatic Volume in Japanese Prostate Cancer Patients with Slightly Elevated Serum PSA Levels
Introduction
Materials And Methods
Results
Correlation Between Total PSA and Age or Prostatic Volume and PPV
Suitable Reference Values According to Prostatic Volume
Discussion
Acknowledgments
References
Determination of Reference Values for Total PSA, F/T and PSAD According to Prostatic Volume in Japanese Prostate Cancer Patients with Slightly Elevated Serum PSA Levels
Background: For screening prostate cancer (CAP) using prostate-specific antigen (PSA), the indications of biopsies for patients showing slightly elevated PSA values are still controversial. Furthermore, the dependence of total PSA, free-to-total PSA ratio (F/T) and PSA density (PSAD) on prostatic volume in gray zone cases is unclear. Methods: By analyzing 1913 patients with a serum total PSA ranging from 2.0 to 20 ng/ml, we evaluated the correlation between total PSA and age or prostatic volume and positive predictive value (PPV) in each range for total PSA, age and prostatic volume. Then suitable reference values for total PSA, F/T and PSAD were decided according to prostatic volume. Results: There was no close correlation between PSA and age or volume. The PPV was high in the group with a prostatic volume of 10-30 ml. Prostatic volume was categorized into three groups, <20, 20-40 and [ge]40 ml, and reference values for obtaining a sensitivity of 90% were proposed. The reference values of total PSA and PSAD were lowered and that of F/T was raised with increase in prostatic volume. The specificity was very low for the [ge]40 ml group. The highest specificity of 36% in PSAD was obtained for the <20 ml group. Conclusion: The reference values for total PSA, F/T and PSAD must be changed according to prostatic volume in order to maintain a sufficient diagnostic sensitivity of CAP. Of these parameters, PSAD showed a high specificity in the group with a prostatic volume of <40 ml.
Key words: prostate cancer - PSA gray zone - F/T - PSAD - prostatic volume
INTRODUCTION
The determination of serum prostate-specific antigen (PSA) has become an essential tumor marker for the diagnosis, evaluation of treatment and follow-up of patients with prostate cancer (CAP) (1-3). Owing to the biological nature of PSA, its serum values may be elevated even in benign prostatic diseases such as acute prostatitis and benign prostatic hyperplasia (BPH). Therefore, the screening and diagnosis of CAP is not accurate for cases showing slightly elevated serum PSA values.
In attempts to improve diagnostic specificity while maintaining sensitivity, age-specific PSA reference values (4), PSA velocity (5), detection of PSA molecular forms and calculation of the ratio in total PSA (6-12) and PSA density (PSAD) (13,14) have been studied. Of these, the free-to-total PSA ratio (F/T) and PSAD are considered to be promising tools. However, a suitable reference value of F/T has not yet been decided even in theUSA or Europe owing to the lack of evaluation of identity among free PSA assays.
In Japan, it has recently been reported that the detection rate of CAP differs according to the size of the prostate (9,15,16). Therefore, we evaluated positive predictive values according to the total PSA, patient's age and prostatic volume in the group having serum PSA values of 2.0-20 ng/ml. We then proposed suitable reference values for total PSA, F/T and PSAD according to the volume.
MATERIALS AND METHODS
Patients with a total PSA value of 2.0-20 ng/ml were enrolled in this study from the Institutes participating in the National Research Project on the Efficacy of Mass Screening for Prostate Cancer, Japan, funded by the Cancer Research of the Japanese Ministry of Health and Welfare (Project Director Professor Hiroki Watanabe) (n = 577) and data for patients showing the same range of total PSA as mentioned above were collected from the 13th Tokyo Prostate Symposium held in December 1997 (n = 1336).
Serum total PSA values were mainly detected by Tandem-R PSA (n = 479) and compatible methods such as Abbott, Eiken, Tosoh and CanAg (n = 892). Those measured by other methods such as MARKIT-M PA (n = 541) were converted to those of Tandem-R using the equation proposed by the Japanese Urological Association (17). Serum free PSA values were detected by Hybritech in 479 cases and those by other methods (n = 479) were converted to those of Hybritech using an equation that we reported previously (18). The prostates were examined by transrectal ultrasonography (TRUS) and their volumes were calculated as a rotary ellipse. All patients were histologically confirmed by six-sectant systematic core needle biopsy or transurethral resection of the prostate. Serum detection of total PSA and free PSA was performed before treatment.
Of the total of 1913 patients enrolled in this study, 491 were of CAP and the other 1422 had no evidence of malignancy of the prostate gland (NEM). BPH was mainly seen in the NEM cases. The clinical stage of CAP according to the criteria of the Japanese Urological Association (19) was stage A in 47, B in 267, C in 81, D in 73 and unclear in 23. Histopathological grade in CAP (19) was well differentiated adenocarcinoma in 176, moderately differentiated adenocarcinoma in 208, poorly differentiated adenocarcinoma in 62 and unclassified adenocarcinoma in 45. Of these cases, a total of 714 patients (CAP 184 and NEM 530) could be evaluated for total PSA, F/T and PSA as determined simultaneously.
We evaluated the correlation between total PSA values and age or prostatic volume and positive predictive values (PPV) of CAP in terms of each serum PSA range, age and prostatic volume. We then proposed suitable reference values for total PSA, F/T and PSAD according to the prostatic volume. All statistical analyses were performed by SPSS (20) and a difference with a p value by Student's t-test of <0.05 was regarded as significant.
RESULTS
Correlation Between Total PSA and Age or Prostatic Volume and PPV
Total PSA values and other parameters such as patient's age at diagnosis, prostatic volume, free PSA, F/T and PSAD were compared according to total PSA in the range 2-20 ng/ml (Table 1). In this range of total PSA, the mean total PSA values were 9.61 ng/ml in patients with CAP and 6.51 ng/ml in those with NEM, with a significant difference. Although there was overlap in the ranges, marked significant differences in the age, prostatic volume, F/T and PSAD values between patients with CAP and NEM were observed. Only free PSA values showed no significant difference with slightly lower values in CAP patients than in NEM.
Table 1. Statistical comparison of each parameter in patients with a total PSA of 2-20 ng/ml
| Parameter | CAP | NEM | p-Value | ||||||
| n | Range | Mean ± SD | Median | n | Range | Mean ± SD | Median | ||
| Age (years) | 427 | 48-94 | 71.4 ± 7.64 | 71 | 1135 | 24-95 | 69.4 ± 8.43 | 69 | 5.00 × 10-6 |
| Volume (ml) | 327 | 4.0-221 | 30.9 ± 19.2 | 27.4 | 1063 | 2.5-152 | 38.9 ± 20.8 | 35.1 | 1.70 × 10-10 |
| Total PSA (ng/ml) | 491 | 2.1-20 | 9.61 ± 4.71 | 9 | 1422 | 2.1-20 | 6.51 ± 3.9 | 5.4 | 2.40 × 10-35 |
| Free PSA (ng/ml) | 276 | 0.10-9.82 | 1.45 ± 1.35 | 1.01 | 682 | 0.07-25.5 | 1.53 ± 1.62 | 1.16 | 0.4295 |
| F/T | 276 | 0.012-1.0 | 0.153 ± 0.122 | 0.12 | 682 | 0.013-1.7 | 0.222 ± 0.127 | 0.201 | 3.00 × 10-14 |
| PSAD | 327 | 0.026-2.17 | 0.397 ± 0.270 | 0.336 | 1063 | 0.015-3.16 | 0.220 ± 0.207 | 0.163 | 7.00 × 10-24 |
The PPV according to total PSA range is shown in Fig. 1. A total of 491 patients with CAP were detected among 1913 patients examined, PPV being 25.7% in total. The PPV increased from 12.5% in the group with a total PSA of 2.0-4.0 ng/ml to 46.5% in the group with a total PSA >10 ng/ml. The correlation between total PSA values and patient's age or prostatic volume was compared. In the range 2.0-20 ng/ml, total PSA was not correlated with age (Fig. 2). The linear regression lines in both CAP and NEM were almost parallel to the x-axis and the correlation coefficient in both groups was <0.02. The PPV showed an extremely low incidence in the age group of <55 years old; only three of the 427 CAP cases were distributed in this range and the detection rate was high in the age group [ge]90 years. In the age group 55-89 years PPV showed a mild increasing trend according to age (Fig. 3). On the other hand, a slightly positive correlation between total PSA value and prostatic volume was observed in NEM patients (Fig. 4). In NEM patients, an increase in volume of 1 ml means an elevation in total PSA value of 0.86 ng/ml. The PPV according to the volume is presented in Fig. 5. The CAP cases showed a high frequency distribution in the group with a prostatic volume of 10-30 ml. In patients with a prostatic volume [ge]60 ml, the PPV was about one third of that in patients with a value of 10-20 ml and accounted for only 17 of the 327 CAP patients.
Figure 1. Positive predictive values in detection of prostate cancer according to total PSA values of 2-20 ng/ml. Positive predictive values (PPV) tended to become high with increase in total PSA values.
Figure 2. Correlation between total PSA and age in patients with a total PSA of 2-20 ng/ml. There was no positive correlation between PSA values and age for either CAP or NEM cases. The linear regression lines were y = 0.0359x + 71.077 (r2 = 0.0005) in CAP (n = 427) and y = 0.2252x + 67.873 (r2 = 0.0108) in NEM (n = 1135).
Figure 3. Positive predictive values in detection of prostate cancer according to patient's age. In the age group 55-89 years, there was no marked age-related tendency in PPV.
Figure 4. Correlation between total PSA and prostatic volume in patients with a total PSA of 2-20 ng/ml. There was a slight positive correlation between total PSA and prostatic volume in NEM cases. The linear regression lines were y = 0.281x + 28.132 (r2 = 0.0044) in CAP (n = 327) and y = 1.162x + 31.115 (r2 = 0.0478) in NEM (n = 1063).
Figure 5. Positive predictive values in detection of prostate cancer according to prostatic volume. The group with a prostatic volume of 10-30 ml had a high incidence of prostate cancer.
Suitable Reference Values According to Prostatic Volume
The specificity of total PSA, F/T and PSAD and also suitable cut-off values for obtaining a diagnostic sensitivity of 90% were compared in the 714 patients (184 in CAP and 530 in NEM) in which all parameters were simultaneously evaluable (Table 2). Total PSA showed a specificity of 20.4%. Therefore, biopsy could be avoided in17.6% of the cases. Similarly, F/T and PSAD showed specificities of 17.9 and 28.5%, respectively. The fixed cut-off values were calculated to be 3.8 ng/ml for total PSA, 0.31 for F/T and 0.126 for PSAD. The PSAD test gave the best result; biopsy could be avoidable in 23.7% of the cases.
Table 2. Comparison of specificity of total PSA, F/T and PSAD in patients with serum PSA values of 2-20 ng/ml for maintaining a diagnostic sensitivity of 90%
| Parameter | Total PSA | F/T | PSAD |
| Reference values | 3.8 ng/ml | 0.31 | 0.126 |
| Sensitivity (%) | 90.2 | 90.2 | 90.2 |
| Specificity (%) | 20.4 | 17.9 | 28.5 |
| No. of patients with avoidable biopsy | 126 | 113 | 169 |
| % | 17.6 | 15.8 | 23.7 |
After the prostatic volume had been categorized into three groups, <20, 20-40 and [ge]40 ml, suitable reference values for total PSA, F/T and PSAD were proposed for obtaining a sensitivity of 90% and the specificity in each category was also calculated ( Table 3). The reference values for total PSA fluctuated from 4.1 to 3.7 and then to 4.7 ng/ml with increase in volume. The reference values for F/T increased from 0.248 to 0.459 as the volume increased. The reference values for PSAD decreased from 0.300 to 0.0624 with increase in the volume. As shown in Fig. 6, the specificity tended to be low in larger prostatic glands. PSAD showed the best specificity in patients with a prostatic volume of <40 ml and total PSA in those with a volume of [ge]40 ml.
Figure 6. Comparison of specificity for obtaininga sensitivity of 90%. When the prostatic volume was >40 ml, the specificity was extremely low for total PSA, F/T and PSAD.
Table 3. Comparison of reference values and specificities of total PSA, F/T and PSAD according to prostatic volume range in patients with serum PSA values of 2-20 ng/ml for maintaining a diagnostic sensitivity of 90%
| Prostatic volume (ml) | Sensitivity (%) | No. of patients | Total PSA | F/T | PSAD | |
| <20 | 91 | CAP: 45 | Reference value | 4.1 ng/ml | 0.248 | 0.3 |
| NEM: 86 | Specificity (%) | 24 | 24 | 36 | ||
| 20-40 | 91 | CAP: 98 | Reference value | 3.7 ng/ml | 0.278 | 0.133 |
| NEM: 246 | Specificity (%) | 22.4 | 23.2 | 27.2 | ||
| [ge]40 | 90 | CAP: 41 | Reference value | 4.7 ng/ml | 0.459 | 0.0624 |
| NEM: 196 | Specificity (%) | 18.9 | 5.6 | 12.8 |
DISCUSSION
There is a limitation to the application of serum PSA determination for differential diagnosis between CAP and NEM in those patients showing slightly elevated serum PSA values. This is natural because PSA itself is not a cancer-specific tumor marker. PSA is a prostate-tissue specific proteinase (21). Several attempts have been made to improve the diagnostic accuracy in patients with slightly elevated PSA levels, especially the evaluation of specificity to obtain a sufficient diagnostic sensitivity. Of these, F/T (10,12) and PSAD (13,15) seem to show a relatively good specificity. In many reports from Western countries, the specificity using F/T in patients with a serum PSA of 4-10 ng/ml was about 30% (6,10-12). However, a uniform cut-off value for F/T has not been decided; according to Brawer et al. (22), one reason is that the difference in free PSA values among assays leads to different F/T values.
In Japan, the specificity of F/T in patients with PSA of 4-10 ng/ml was reported by Kuriyama et al. (15) to be 19.6% with 90.7% sensitivity. In another study (8) we found F/T to be the most effective for the differential diagnosis of CAP from NEM in a group of higher PSA levels in the gray zone; the specificity increased to 60.7% in the 5-10 ng/ml group from 18.8% in the 4-10 ng/ml group with the same sensitivity of 90%. Furthermore, the specificity of PSAD showed better results than F/T (15). Akimoto et al. (16) also found a closer relation between PSA and prostatic volume than patient's age in Japanese mass screening subjects. They recommended a PSA reference range calculated with a stratified prostatic volume. Therefore, we re-evaluated the relationship between total PSA and patient's age or prostatic volume in a large number of cases. To our knowledge, this is the largest study series in Japan to date. If the prostatic volume or patient's age can be shown to work as an independent indicator for the prediction of CAP in the group with a total PSA of 2.0-20 ng/ml, we can propose a kind of volume-specific or age-specific reference value for total PSA, F/T and PSAD. In each condition, various specificities maintaining a sensitivity of 90% were also obtained because many clinicians expect at least this level of diagnostic sensitivity in this group. Catalona et al. also compared the specificity of F/T at the point showing a diagnostic sensitivity of 95 or 90% (10). Furthermore, in our studies, since the total PSA and free PSA values were either detected by the Tandem-R series or converted by using the equation obtained in our previous studies (17,18), the difference derived from assay methods for total and free PSA could be ignored.
The correlation between total PSA and patient's age was extremely low in the group with a total PSA value of 2-20 ng/ml, as Akimoto et al. reported (16). There was no difference between patients with CAP and NEM. Total PSA and prostatic volume in NEM cases showed a mild positive correlation. The PPV was 12.5% in the group with a PSA value of 2.0-4.0 ng/ml and gradually increased, as expected, with elevation of the total PSA value. When the total PSA was [ge]6.0 ng/ml, the PPV exceeded 20%; this level might be acceptable to patients for biopsy (23). The PPV showed a slight increase with age. The PPV was low in patients aged <55 and high in those of [ge]90 years. However, the incidence of CAP was low in both age groups: only three and five patients, respectively, out of 427 patients with CAP in this series. In those aged between 55 and 90 years, the PPV could be classified into two groups: about 25% in those <70 years and about 30% in those in their seventies or eighties. The PPV differed greatly with prostatic volume. The patients with a prostatic volume of 10-30 ml showed a higher PPV than the average ,which was 23.5%. Imai et al. (9) described the possibility of CAP being very low in patients with a total PSA of <10 ng/ml and a prostatic volume of <40 ml. In our series, the number of CAP patients with a prostate volume of >40 ml was 64 of the 327 (19.6%) whose prostate volume was examined.
The overall specificities calculated in 714 patients who were simultaneously evaluated for total PSA, F/T and PSAD were 20.4, 17.9 and 28.5%, respectively. The usefulness of F/T was less than that of total PSA for maintaining a diagnostic sensitivity of >90%. Then, suitable reference values were calculated according to the prostatic volume. The reference values for total PSA fluctuated. However, when calculated on 327 CAP patients and 1063 NEM patients who had been subjected to simultaneous determination of total PSA and prostatic volume, regardless of free PSA determination, the reference values decreased as the volume increased: 4.1 ng/ml in the <20 ml group, 3.9 in the 20-40 ml group and 3.5 ng/ml in the [ge]40 ml group. The specificities were 32.5, 34.2 and 15%, respectively. Similarly, the reference values and specificities for PSAD were 0.257 and 27.6% in the <20 ml group, 0.149 and 44.3% in the 20-40 ml group and 0.063 and 14.3% in the [ge]40 ml group.
Hence the reference values must be changed according to prostatic volume, and with an increase in prostatic volume to keep the sensitivity sufficiently high, a lower total PSA and PSAD and a higher F/T. Moreover, it must be noted that the specificity in the [ge]40 ml group was still low regardless of the method. In a large prostate with especially small cancer foci, the diagnosis of CAP by SSB may be underestimated, which leads to a low specificity using these parameters. Although CAP cases with a large prostate are not common, a large prostate may cause worse symptoms of lower urinary obstruction than expected and should be checked by urologists with other diagnostic imaging methods.
The prostatic volume is neither easy to measure, especially by TRUS, nor as objective as serum total or free PSA detection. Moreover, a suitable detection method has not yet been decided. Transabdominal ultrasonography (TAUS) of the prostate may be used alternatively to TRUS only for volume detection (14,16). This may be easy especially in mass screening. Although it will be complicated and less cost-effective to detect prostatic volume than PSA determination, PSAD showed a 5-10% higher specificity than total PSA in the group with a prostatic volume of <40 ml. This difference is large. For every million people with a PSA gray zone in mass screening, we may avoid 50,000-100,000 unnecessary and harmful biopsies.
In conclusion, the reference values for total PSA, F/T and PSAD must be changed according to the prostatic volume to maintain a high diagnostic sensitivity. Of these parameters, PSAD showed the highest specificity in patients with a prostatic volume of <40 ml. The specificity of <19% of total PSA was low in the group with a larger prostatic volume ([ge]40 ml).
Acknowledgments
The authors express their sincere appreciation to researchers participating in the National Research Project on the Efficacy of Mass Screening for Prostate Cancer, Japan, funded by the Cancer Research of the Japanese Ministry of Health and Welfare, and the 13th Tokyo Prostate Symposium for providing cases and medical records.
References
Received August 3, 1999; accepted September 27, 1999For reprints and all correspondence: Manabu Kuriyama, Department of Urology, Gifu University School of Medicine, 40 Tsukasa-machi, Gifu 500-8705, Japan. E-mail: mkuriyam-gif{at}umin.ac.jp
Abbreviations: PSA, prostate-specific antigen; CAP, prostate cancer; NEM, no evidence of malignancy of the prostate; BPH, benign prostatic hyperplasia; F/T, free-to-total PSA ratio; PSAD, PSA density; PPV, positive predictive value
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© 1999 Foundation for the Promotion of Cancer Research.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||