0022-5347 /92/14 73-0817$03.00/0 Vol. 147, 817-821, March 1992
THE JOURNAL OF UROLOGY
Copyright© 1992 by AMERICAN UROLOGICAL ASSOC!AT!ON, INC.
Printed in US.A.
THE USE OF PROSTATE SPECIFIC ANTIGEN DENSITY TO ENHANCE THE PREDICTIVE VALUE OF INTERMEDIATE LEVELS OF SERUM PROSTATE SPECIFIC ANTIGEN MITCHELL C. BENSON,* IHN SEONG WHANG, CARL A. OLSSON, DONALD J. McMAHON WILLIAM H. COONER
AND
From the Department of Urology, J. Bentley Squier Urological Clinic, Columbia-Presbyterian Medical Center, Department of Urology and Irving Center for Clinical Research, Columbia University College of Physicians and Surgeons, New York, New York, and University of South Alabama, Mobile, Alabama
ABSTRACT
Prostate specific antigen (PSA) is an extremely valuable tumor marker. However, its use in detection is limited by its low positive and negative predictive values, The ability of serum PSA to distinguish between benign and malignant prostatic conditions is particularly poor in the intermediate range of 4.1 and 10 ng.f ml. by the Hybritech assay. We used transrectal ultrasound determined prostate volumes in a well characterized population of 533 men to form a serum PSA/prostate volume ratio called prostate specific antigen density (PSAD). The prevalence of cancer in the entire population was 18.4%. Discriminant analysis according to negative or positive outcome allowed for the construction of nomograms, which resulted in a PSAD defined cancer risk ranging from 3 to 100%. Predictive value nomograms created from PSAD may allow for a more individualized approach to evaluation of patients with intermediate levels of Hybritech serum PSA. KEY WORDS:
adenocarcinoma; prostatic neoplasms; hypertrophy; antigens, differentiation
The discovery of prostate specific antigen (PSA) in 1979 by Wang et al marked the beginning of a new era in the management of prostate cancer. 1 PSA is a 34,000 dalton serine protease that can be detected in normal prostate epithelium, benign prostatic hypertrophy (BPH), malignant prostate epithelium, prostatic fluid and serum. It is not detectable in significant quantities in other human tissues. 2 It is this specificity that makes it a valuable tumor marker in the longitudinal assessment of patients with prostate cancer. 3 - 13 Its value in screening is more controversial since PSA is not prostate cancer specific. 3· 4· 6· 7 At the time of this study the Yang assay and the Hybritech assay were available to detect and quantitate PSA. 4 • 8 The Yang assay uses a polyclonal antibody, while the Hybritech assay uses a monoclonal antibody. The Hybritech assay was used in this report. Hybritech PSA is least specific when it is between 4.1 and 10 ng./ml. When PSA is 4 ng./ml. or less there is a low prevalence of prostate cancer (1.4%), 5 and when it is greater than 10 ng./mL there is a high prevalence of prostate cancer. Because a relatively high percentage of patients with BPH have PSA values between 4.1 and 10 ng./ml., these patients pose a clinical dilemma. To assign these cases correctly transrectal prostatic ultrasonography and guided biopsy are performed. 13 - 15 To identify more precisely patients with prostate cancer among patients with PSA levels of 4.1 to 10 ng./ml., we used a concept of prostate specific antigen density (PSAD), which we first reported in 1989. 16 PSAD is defined as serum PSA divided by the volume of the prostate. This study demonstrates that PSAD is better able to distinguish patients with prostatic cancer from those with BPH than PSA alone. MATERIALS AND METHODS
Between January 1987 and July 1990, 595 men 50 to 89 years old with PSA values greater than 4 ng./ml. and less than or equal to 10 ng./ml. (Hybritech assay) were examined by digital rectal examination and prostate ultrasonography. A portion of Accepted for publication October 18, 1991. * Requests for reprints: 630 W. 168th St., New York, New York 10032.
this population has been reported previously. 13 • 14 All patients were referred for urological evaluation and, therefore, the study group does not represent a screening population. To minimize the subjectivity of digital rectal examination 2 urologists had to agree that the prostate gland was palpably normal before the patient was considered to have a normal gland. If either of the urologists palpated an abnormality of the gland, the patient was assigned to the abnormal digital rectal examination category. PSA levels were determined by the Hybritech assay at least 1 week after any rectal or urethral manipulation to avoid false elevations of PSA. Transrectal prostate ultrasonography was performed in real-time using a Brue! and Kjaer Model 1846 with 7 MHz. transducers. Every gland was visualized in the transaxial and longitudinal planes. All sonography and biopsies were performed by l author (W. H. C.) using a consistent method that has been reported previously. 13 Hypoechoicism unexplained by other factors, such as normal anatomical areas, vascular structures, cysts or artifacts, was the sole transrectal ultrasonography criterion used to determine the need for biopsy. No patient in this group underwent random biopsies. All biopsies were performed under sonographic guidance transrectally using either the Menghini needlet or the Biopty needle.:j: The volume of the prostate was calculated using the prolate ellipse formula. Volume equals 0.52 (L x W X H), where L is length obtained from the longitudinal sonographic view, Wis width obtained from the transaxial view and H is height obtained transaxially. This technique is adaptable to any transrectal ultrasonography assessment of the prostate that images the prostate in longitudinal and transaxial views. The 595 patients were divided into cancer and noncancer groups. Sixty-two patients had either an abnormal digital rectal examination or abnormal ultrasonography but did not undergo biopsy. These patients were excluded from analysis since the etiology of the abnormalities was not determined. Thus, the final population analyzed was 533 patients. The cancer group included 98 patients with biopsy proved adenocarcinoma of the prostate. There were 435 patients in whom a diagnosis of cancer was not established. Of these patients 191 had negative biopsies.
817
t Meadox Surgimed, Inc., Oakland, New Jersey. :j: Bard Urological, Covington, Georgia.
818
BENSON AND ASSOCIATES
This group was analyzed as an independent control. There were 244 patients who had no indication for biopsy, normal rectal examination and normal transrectal ultrasonography. This group was analyzed independently and also combined with the negative biopsy group to form a no clinical evidence of prostate cancer group of 435 patients. It was assumed that these patients were free of cancer although it is possible that there were patients with prostate cancer among them. A PSA between 4.1 and 10 was not considered an indication for biopsy in the absence of other findings. Statistical significance was determined by t tests performed using the computer based Number Cruncher Statistical System. PSAD values from patients who did and did not undergo biopsy were subjected to a discriminant analysis (PROC DISCRIM) aeeording to positive or ·negative outcome. Normal- -density estimates and equal variances were used. Four models were tested which differed in the assumed prevalence of cancer in the population and by whether or not cases with no indication for prostate biopsy (normal rectal examination and transrectal ultrasonography) were considered true negatives. Models 1 and 2 assumed prevalences of 34% (the frequency of cancer for the population undergoing biopsy) and 18.4% (the frequency of cancer if patients not undergoing biopsy were assumed to be free of disease), and only considered patients who underwent biopsy. Models 3 and 4 assumed similar prevalences and included as negative patients who had no indication for biopsy. If a patient had a negative initial biopsy with a positive subsequent biopsy within 1 year he was considered to have had a falsely negative biopsy initially, and for the purpose of this analysis was considered to have had cancer at the time of the initial biopsy. There were 6 patients who were included in this category. If a patient had a change in diagnosis from benign to malignant more than 1 year after the initial biopsy, the initial biopsy was analyzed as a true negative. For patients with multiple PSA determinations the PSA closest in time to transrectal ultrasonography was used for determination of PSAD. RESULTS
There were 533 patients who had PSA values greater than 4 ng./ml. and less than or equal to 10 ng./ml. by the Hybritech assay. Among them 289 underwent transrectal biopsy under sonographic guidance (biopsy rate 54%), of whom 98 (34%) had prostate cancer found on biopsy, while in 191 (66%) no cancer was identified. There were 244 patients who did not undergo biopsy. Table 1 lists the characteristics of the entire population studied. The negative biopsy and no biopsy groups are shown independently. The mean PSA values were 7.0, 6.8 and 6.1 ng./ ml. for the positive biopsy, negative biopsy and no biopsy groups, respectively. The volumes were 28.9, 40.1 and 44.0 cc, respectively. The mean PSAD values were 0.297, 0.208 and 0.172, respectively. The volumes of the no biopsy and negative biopsy groups are similar. The patients with biopsy proved prostate cancer and biopsy proved nonmalignant disease are analyzed in table 1. The prevalence of cancer in this population was 34%. This analysis reveals that the mean PSAD values are TABLE
No.
Pos. biopsy Neg. biopsy No biopsy
Mean PSA (ng./ml.)
statistically significantly different between the positive biopsy and the negative biopsy groups (p
THE JOURNAL OF UROLOGY
Copyright© 1992 by AMERICAN UROLOGICAL ASSOC!AT!ON, INC.
Printed in US.A.
THE USE OF PROSTATE SPECIFIC ANTIGEN DENSITY TO ENHANCE THE PREDICTIVE VALUE OF INTERMEDIATE LEVELS OF SERUM PROSTATE SPECIFIC ANTIGEN MITCHELL C. BENSON,* IHN SEONG WHANG, CARL A. OLSSON, DONALD J. McMAHON WILLIAM H. COONER
AND
From the Department of Urology, J. Bentley Squier Urological Clinic, Columbia-Presbyterian Medical Center, Department of Urology and Irving Center for Clinical Research, Columbia University College of Physicians and Surgeons, New York, New York, and University of South Alabama, Mobile, Alabama
ABSTRACT
Prostate specific antigen (PSA) is an extremely valuable tumor marker. However, its use in detection is limited by its low positive and negative predictive values, The ability of serum PSA to distinguish between benign and malignant prostatic conditions is particularly poor in the intermediate range of 4.1 and 10 ng.f ml. by the Hybritech assay. We used transrectal ultrasound determined prostate volumes in a well characterized population of 533 men to form a serum PSA/prostate volume ratio called prostate specific antigen density (PSAD). The prevalence of cancer in the entire population was 18.4%. Discriminant analysis according to negative or positive outcome allowed for the construction of nomograms, which resulted in a PSAD defined cancer risk ranging from 3 to 100%. Predictive value nomograms created from PSAD may allow for a more individualized approach to evaluation of patients with intermediate levels of Hybritech serum PSA. KEY WORDS:
adenocarcinoma; prostatic neoplasms; hypertrophy; antigens, differentiation
The discovery of prostate specific antigen (PSA) in 1979 by Wang et al marked the beginning of a new era in the management of prostate cancer. 1 PSA is a 34,000 dalton serine protease that can be detected in normal prostate epithelium, benign prostatic hypertrophy (BPH), malignant prostate epithelium, prostatic fluid and serum. It is not detectable in significant quantities in other human tissues. 2 It is this specificity that makes it a valuable tumor marker in the longitudinal assessment of patients with prostate cancer. 3 - 13 Its value in screening is more controversial since PSA is not prostate cancer specific. 3· 4· 6· 7 At the time of this study the Yang assay and the Hybritech assay were available to detect and quantitate PSA. 4 • 8 The Yang assay uses a polyclonal antibody, while the Hybritech assay uses a monoclonal antibody. The Hybritech assay was used in this report. Hybritech PSA is least specific when it is between 4.1 and 10 ng./ml. When PSA is 4 ng./ml. or less there is a low prevalence of prostate cancer (1.4%), 5 and when it is greater than 10 ng./mL there is a high prevalence of prostate cancer. Because a relatively high percentage of patients with BPH have PSA values between 4.1 and 10 ng./ml., these patients pose a clinical dilemma. To assign these cases correctly transrectal prostatic ultrasonography and guided biopsy are performed. 13 - 15 To identify more precisely patients with prostate cancer among patients with PSA levels of 4.1 to 10 ng./ml., we used a concept of prostate specific antigen density (PSAD), which we first reported in 1989. 16 PSAD is defined as serum PSA divided by the volume of the prostate. This study demonstrates that PSAD is better able to distinguish patients with prostatic cancer from those with BPH than PSA alone. MATERIALS AND METHODS
Between January 1987 and July 1990, 595 men 50 to 89 years old with PSA values greater than 4 ng./ml. and less than or equal to 10 ng./ml. (Hybritech assay) were examined by digital rectal examination and prostate ultrasonography. A portion of Accepted for publication October 18, 1991. * Requests for reprints: 630 W. 168th St., New York, New York 10032.
this population has been reported previously. 13 • 14 All patients were referred for urological evaluation and, therefore, the study group does not represent a screening population. To minimize the subjectivity of digital rectal examination 2 urologists had to agree that the prostate gland was palpably normal before the patient was considered to have a normal gland. If either of the urologists palpated an abnormality of the gland, the patient was assigned to the abnormal digital rectal examination category. PSA levels were determined by the Hybritech assay at least 1 week after any rectal or urethral manipulation to avoid false elevations of PSA. Transrectal prostate ultrasonography was performed in real-time using a Brue! and Kjaer Model 1846 with 7 MHz. transducers. Every gland was visualized in the transaxial and longitudinal planes. All sonography and biopsies were performed by l author (W. H. C.) using a consistent method that has been reported previously. 13 Hypoechoicism unexplained by other factors, such as normal anatomical areas, vascular structures, cysts or artifacts, was the sole transrectal ultrasonography criterion used to determine the need for biopsy. No patient in this group underwent random biopsies. All biopsies were performed under sonographic guidance transrectally using either the Menghini needlet or the Biopty needle.:j: The volume of the prostate was calculated using the prolate ellipse formula. Volume equals 0.52 (L x W X H), where L is length obtained from the longitudinal sonographic view, Wis width obtained from the transaxial view and H is height obtained transaxially. This technique is adaptable to any transrectal ultrasonography assessment of the prostate that images the prostate in longitudinal and transaxial views. The 595 patients were divided into cancer and noncancer groups. Sixty-two patients had either an abnormal digital rectal examination or abnormal ultrasonography but did not undergo biopsy. These patients were excluded from analysis since the etiology of the abnormalities was not determined. Thus, the final population analyzed was 533 patients. The cancer group included 98 patients with biopsy proved adenocarcinoma of the prostate. There were 435 patients in whom a diagnosis of cancer was not established. Of these patients 191 had negative biopsies.
817
t Meadox Surgimed, Inc., Oakland, New Jersey. :j: Bard Urological, Covington, Georgia.
818
BENSON AND ASSOCIATES
This group was analyzed as an independent control. There were 244 patients who had no indication for biopsy, normal rectal examination and normal transrectal ultrasonography. This group was analyzed independently and also combined with the negative biopsy group to form a no clinical evidence of prostate cancer group of 435 patients. It was assumed that these patients were free of cancer although it is possible that there were patients with prostate cancer among them. A PSA between 4.1 and 10 was not considered an indication for biopsy in the absence of other findings. Statistical significance was determined by t tests performed using the computer based Number Cruncher Statistical System. PSAD values from patients who did and did not undergo biopsy were subjected to a discriminant analysis (PROC DISCRIM) aeeording to positive or ·negative outcome. Normal- -density estimates and equal variances were used. Four models were tested which differed in the assumed prevalence of cancer in the population and by whether or not cases with no indication for prostate biopsy (normal rectal examination and transrectal ultrasonography) were considered true negatives. Models 1 and 2 assumed prevalences of 34% (the frequency of cancer for the population undergoing biopsy) and 18.4% (the frequency of cancer if patients not undergoing biopsy were assumed to be free of disease), and only considered patients who underwent biopsy. Models 3 and 4 assumed similar prevalences and included as negative patients who had no indication for biopsy. If a patient had a negative initial biopsy with a positive subsequent biopsy within 1 year he was considered to have had a falsely negative biopsy initially, and for the purpose of this analysis was considered to have had cancer at the time of the initial biopsy. There were 6 patients who were included in this category. If a patient had a change in diagnosis from benign to malignant more than 1 year after the initial biopsy, the initial biopsy was analyzed as a true negative. For patients with multiple PSA determinations the PSA closest in time to transrectal ultrasonography was used for determination of PSAD. RESULTS
There were 533 patients who had PSA values greater than 4 ng./ml. and less than or equal to 10 ng./ml. by the Hybritech assay. Among them 289 underwent transrectal biopsy under sonographic guidance (biopsy rate 54%), of whom 98 (34%) had prostate cancer found on biopsy, while in 191 (66%) no cancer was identified. There were 244 patients who did not undergo biopsy. Table 1 lists the characteristics of the entire population studied. The negative biopsy and no biopsy groups are shown independently. The mean PSA values were 7.0, 6.8 and 6.1 ng./ ml. for the positive biopsy, negative biopsy and no biopsy groups, respectively. The volumes were 28.9, 40.1 and 44.0 cc, respectively. The mean PSAD values were 0.297, 0.208 and 0.172, respectively. The volumes of the no biopsy and negative biopsy groups are similar. The patients with biopsy proved prostate cancer and biopsy proved nonmalignant disease are analyzed in table 1. The prevalence of cancer in this population was 34%. This analysis reveals that the mean PSAD values are TABLE
No.
Pos. biopsy Neg. biopsy No biopsy
Mean PSA (ng./ml.)
statistically significantly different between the positive biopsy and the negative biopsy groups (p
