Application to Clinical Study Results
The age-specific reference ranges established by Oesterling's group[25] have been applied retrospectively to a number of studies and have substantiated the contention that this PSA index increases sensitivity among younger men and specificity among older men. In one study[38] of 2,998 men more than 60 years of age from a single urologic practice, age-specific reference ranges decreased sensitivity by 9% but increased specificity by 11% and positive predictive value by 5%. If these reference ranges had been used, 92 biopsies (5.5%) would have been avoided, and only 19 (.6%) cancers would not have been detected. However, of these 19 cancers, 13 were detected in men ³ 70 years of age and 18 were unlikely to be of clinical consequence, based on organ confinement, tumor volume, and Gleason grade.
Partin and colleagues[39] reviewed the records of 4,597 men with clinically localized prostate cancer from the extensive Johns Hopkins surgical series. Using the standard reference PSA value of 4.0 ng/mL alone, they found that 82% of tumors would have been detected; age-specific reference ranges would have detected 78%. Age-specific reference ranges would have found 18% more cancers in men less than 60 years of age and 22% fewer cancers in men 60 years of age or more.
In men less than 60 years of age, 74 cancers would have been detected by age-specific reference ranges but not by the standard reference. Moreover, of these 74 cancers, 60 (81%) had a favorable pathologic status: either organ confinement or capsular penetration with a Gleason grade of less than 7. In men 60 years of age or more, 252 tumors would not have been detected, but of these tumors, 76% (191) had a favorable pathologic status. This study concluded that age-specific reference ranges are most useful in men 60 years of age or more with T1c prostate cancer (ie, they increase specificity) because 95% of the "missed" tumors among older men had a favorable pathologic status.
Another study, apparently based on results from early detection programs, concurred that age-specific reference ranges increased sensitivity in detecting prostate cancer in younger men, who would be more likely to benefit from treatment, while decreasing the biopsy rate in older patients.[40] These investigators concluded that age-specific reference ranges would be most valuable for patients more than 70 years of age, among whom 22% would have been spared transrectal ultrasonography with biopsy. Similarly, Lankford and colleagues[41] found a significant association between PSA levels and age (r = .33) and a significant improvement in specificity when age-specific reference ranges were applied to men 70 years of age or more , as compared with the standard PSA reference (58.6% and 34.2%, respectively).
Two European studies[42,43] reported results based on the application of age-specific reference ranges to findings from screening populations. Reissigl and associates[42] reported that an 8% increase in biopsies among younger men (in their 40s and 50s) would have occurred with age-specific reference ranges, but this would have effected an 8% increase in the detection of organ-confined cancer. In older men (more than 60 years), 21% fewer biopsies would have been performed and 4% of organ-confined tumors would have been missed. Using age-specific reference ranges and digital rectal examination as indicators for biopsy, Bangma and colleagues[43] reported that a 37% reduction in biopsies would have occurred, with a 12% loss of detected cancers, all of which were nonpalpable and organ-confined .
Although these studies generally support the use of age-specific reference ranges to improve the characteristics of PSA testing, they do not provide sufficient evidence for their universal application. The variability of findings likely reflects the diagnostic capabilities of the clinical investigators, the detection technologies employed, and the different patient populations. One autopsy study confirmed an increased specificity with age-specific reference ranges among men more than 60 years of age,[44] but such studies simply show that cancers are missed and an elevated PSA value (more than 4.0 ng/mL) alone does not indicate cancer.
Criticisms of Age-Specific Ranges
Widespread acceptance of age-specific PSA reference ranges as a valuable index for the effective and efficient detection of prostate cancer has been compromised by negative clinical findings, philosophy, and the continued adequate performance of the standard reference PSA value of 4.0 ng/mL. Age-specific reference ranges have theoretical appeal, but study results are not consistent with regard to the percentage and type of prostate cancers not detected among older men when age-specific reference ranges have been applied retrospectively.
Data from the American Cancer Society's National Prostate Cancer Detection Project showed that age-specific reference ranges increased specificity but at a nonsignificant loss of sensitivity.[45] A PSA value of 4.0 ng/mL alone had a sensitivity of 71.9% and a specificity of 90%; age-specific reference ranges had a sensitivity of 67.3% and a specificity of 90.9%.
A German study concluded that the application of age-specific reference ranges (or any PSA value less than 3.1 ng/mL) does not improve the test's sensitivity among men younger than age 50. Moreover, using a higher PSA value (7.5 ng/mL) for men more than 70 years of age does improve specificity but lowers sensitivity.[46]
Although Oesterling and colleagues[38] and Partin's group[39] each noted that only 5% of cancers undetected with age-specific reference ranges had unfavorable histology, a recent study[47] reported that 60% (9 of 15) of cancers that would have remained undetected by age-specific reference ranges had histologic characteristics qualifying them as large, life-threatening, and clinically significant. All of these cancers were in men younger than age 75.
In another extensive critique of age-specific reference ranges, Catalona and associates[48] concluded that if an age-specific reference of 3.5 ng/mL had been used for men between the ages of 50 and 59, there would have been a 45% increase in biopsies, with only a projected 15% increase in cancer detection. If an age-specific reference of 4.5 ng/mL had been used for men between the ages of 60 and 69, 15% fewer biopsies would have occurred, but 8% of organ-confined tumors would have been missed. Increasing the normal PSA cut-off value to 6.5 ng/mL for men more than 70 years would have resulted in 44% fewer biopsies (70 of 159) and would have missed 47% of detected organ-confined cancers (7 of 15). This study concluded that a PSA value of 4.0 ng/mL need not be altered for older men because the number of biopsies performed for each cancer detected remained constant across age groupings, apparently unaffected by the simultaneously increasing prevalence of BPH and cancer with age.
In light of the fact that both sensitivity and specificity cannot be improved simultaneously, the philosophical issue surrounding age-specific reference ranges is whether sensitivity or specificity should take priority. Advocates of screening and early detection of prostate cancer favor sensitivity over specificity because more clinically important, organ-confined disease will be detected.[49] In fact, it has been proposed that the threshold for normal be lowered from 4.0 to 2.5 ng/mL for everyone.[50] A more provocative proposal has been to ignore PSA values and age-specific reference ranges completely and simply to perform a biopsy on all men at 50 years of age[51]; specificity would be sacrificed at the altar of universal detection.
These and similar proposals may not be implemented because of the relatively satisfactory performance of a single demarcation between normal and elevated PSA levels.[52] The reference for the maximum normal PSA value (4.0 ng/mL) was established with a population mix younger than would be normally tested for prostate cancer: 472 individuals, of whom only 55 were between the ages of 50 and 59 and 19 were 60 years of ageor more.[53] This basis may not reflect the increasing risk of prostate cancer with increasing age, but the objective was to identify a demarcation between normal and elevated PSA levels. Unless another standard becomes more acceptable, the imperfect nature of the PSA test may simply require such a norm; ie, below 4.0 ng/mL to detect more organ-confined cancers regardless of the higher number of false-positive results or above 4.0 ng/mL to detect fewer clinically insignificant cancers while missing more clinically significant cancers.
Variations on the Age-Specific Reference Range Theme
Attempts to complement the value of age-specific reference ranges in the detection of prostate cancer have included consideration of prostatic volume after age-specific reference ranges and proposed volume-referenced PSA values. Meshref and colleagues[54] studied PSA density with age-specific reference ranges. In a clinically referred population (N = 2,429), transrectal ultrasonography was performed on everyone. This experience alone indicated proficiency in establishing comparable estimations of volume and led to the finding that volume is more strongly correlated to PSA value (r = .46) than are PSA values to age (r = .25). Prostate-specific antigen density (which directly relates serum PSA values to prostatic volume), however, showed a weak correlation with age (r = .1).
In multiple regression analysis, prostate volume accounted for 18% of the variation in serum PSA values, whereas age accounted for only an additional 2%. In the patient group with PSA values between the age-specific upper limit of normal and 10.0 ng/mL, the positive rate of biopsy in patients with a PSA density less than .15 was only 1.9% (2/108/315), but in patients with a PSA density more than .15, the positive rate of biopsy was 27.2% (59/217/240). This study concluded: (1) that age-specific reference ranges do not totally account for the effect of prostate volume on serum PSA values and (2) that PSA density can still safely be used to reduce the number of biopsies performed in patients with a normal digital rectal examination and transrectal ultrasonography, as well as a serum PSA level 10.0 ng/mL or less but above the age-specific reference range limit of normal.
Babaian and associates[29] also attempted to address the mixed results of studies of PSA density and age-specific reference ranges. Because the association of glandular volume with PSA values is stronger than that with age, these investigators developed the concept of volume-referenced PSA testing and compared its sensitivity and specificity with other indices.
Volume-referenced PSA testing was more sensitive but less specific than age-specific reference ranges and PSA density; it was only equivalent to PSA testing clinically. Age-specific reference ranges would have prevented more biopsies than volume-referenced testing (39% vs 31%) but would have resulted in the diagnosis of 48% fewer cancers. Receiver-operator characteristic curve analysis demonstrated a better performance for volume-referenced PSA testing and PSA density than for PSA testing or age-specific reference ranges.
In the PSA value range of 4 to 10 ng/mL, volume-referenced testing was superior to PSA testing (P = less than .01) and age-specific reference ranges (P less than .001) and equivalent to PSA density. In this range, age-specific reference ranges missed the fewest cancers (6%) but resulted in the fewest biopsies (12%). However, a key point of this study was that enhanced detection of prostate cancer was inversely associated with increasing volume of the prostate gland and directly related to increasing patient age.[29]
The relationship between age-specific reference ranges and PSA change--PSA velocity--offers another potential PSA index that could improve the test's characteristics. Pearson and colleagues[55] studied annual PSA velocity rates in 727 men (age range, 35 to 92 years) with no history of prostate cancer and a PSA level < 10 ng/mL. The mean PSA velocity and the upper limits of PSA velocity increased significantly with age (P less than .001). Age-specific values for PSA velocity were suggested: less than 50 years of age, 0.125 ng/mL/yr; 50 to 59 years old, 0.15 ng/mL/yr; 60 to 69 years old, 0.3 ng/mL/yr; and 70 years older more, 0.4 ng/mL/yr.[55]
The shortcoming of this study, however, is the particular PSA range of values selected (ie, 4.0 to 10.0 ng/mL) to establish age-specific PSA velocity: Whatever range of PSA values is used to establish "normal" PSA velocity will directly influence the resultant PSA velocity determinations. If, for example, a PSA value of 0.0 to 4.0 ng/mL is used to establish PSA velocity levels, the age-specific rates will be much lower. Again, it becomes a question of whether to emphasize sensitivity or specificity in the detection of prostate cancer by using age-specific PSA velocity values.
