| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1
CNR, Inst. of Clin. Physiol., via Savi 8, 56100 Pisa, Italy;
2
Serv. de Radiopharmacie et de Radioanalyse, Univ. de Lyon, Lyon, France;
a author
for correspondence: fax 0039-50-553461, e-mail zucchell{at}po.ifc.pi.cnr.it
Prostate-specific antigen (PSA) is a glycoprotein that is secreted by the prostate into the seminal fluid. Low concentrations of the protein are normally released into blood, but in prostate cancer (CAP) as well as in a high proportion of subjects with benign prostatic hyperplasia (BPH), serum PSA concentrations frequently increase above normal values (4 µg/L). Therefore, immunoassays measuring serum PSA concentration are routinely carried out to diagnose prostate diseases and to monitor progress of the disease and relapse of CAP after removal of the prostate (1)(2)(3).
A few years ago, PSA was reported to be present in serum in three
different forms. The predominant molecular form is complexed to
1-antichymotrypsin, whereas a minor fraction circulates
in a free noncomplexed form. These two forms are both measured by PSA
assay. Only a very small proportion of PSA circulates bound to
2-macroglobulin; this third form, however, is a
nonimmunoreactive complex.
The free, noncomplexed form of PSA is reported to constitute a minute proportion of the serum PSA in patients with CAP, but to be significantly greater in subjects affected by BPH. On the basis of this observation the simultaneous measurement of total PSA (tPSA) and free PSA (fPSA) has been suggested. The computed ratio fPSA/tPSA is considered a useful tool to better discriminate between BPH subjects and CAP patients and therefore to improve the early diagnosis of CAP (4).
Many immunoassays for fPSA measurement have been developed and are now commercially available. To evaluate the analytical performance of these assays, the international External Quality Assessment (EQA) program "Oncocheck" for tumor markers (AFP, CEA, CA 19–9, CA 15–3, CA 125, tPSA) organized by Service de Radiopharmacie et Radioanalyse, University of Lyon in cooperation with our Institute and Cis BioInternational has been extended to fPSA assay (5)(6). About 300 laboratories participated in the 1996 EQA cycle, assaying tPSA; among these about 70 laboratories also assayed fPSA in control samples.
The most popular methods used by participants in the EQA for fPSA assay were IRMA Hybritech; IRMA Cis, Cis Biointernational; and ICMA Immulite, Diagnostic Products Corp. Each of these methods was used by about 20 laboratories.
Control samples were prepared by diluting a serum pool (tPSA concentration ~2000 µg/L) obtained from patients affected by CAP with normal human serum (tPSA concentration <0.5 µg/L); different dilutions were made to cover the entire assay range.
During the 1996 EQA cycle, 22 control samples (freeze-dried) were distributed and assayed; their average concentrations (consensus mean of all reported results) ranged from 1.99 to 28.1 µg/L for tPSA and from 0.15 to 2.07 µg/L for fPSA. The average between-laboratory agreement (or total variability, CV) of fPSA determinations was 28.0%. This variability was decomposed by ANOVA technique in the between-method and within-method components (7)(8).
The within-method component (an estimate of the precision of the "average" method) was 21.8%, accounting for 60% of variability. This figure indicates that the methods for fPSA assays are affected by poor precision when compared with the within-method precision of tPSA (14.9%, computed from results of the same control samples).
The between-method component (which reflects the systematic differences
in results produced by different methods) was 17.6%, accounting for
the remaining 40% of the total variability. In fact, average fPSA
results produced by the three most popular methods are consistently
different from each other. This last observation is clearly appreciated
from regression analysis reported in Fig. 1
; it can be calculated, from regression equations, that 1 µg/L
fPSA measured by IRMA Hybritech corresponds to 1.22 µg/L of IRMA Cis
and 0.78 µg/L of ICMA Immulite (22% of overestimation and
underestimation, respectively). This scarce agreement indicates poor
relative accuracy of the methods and can be explained both by
differences in antibody specificities and (or) by differences in
calibrators.
|
The precision of the individual fPSA methods was estimated by averaging the CVs of all results produced by the method during the whole EQA cycle for the same control sample (assayed in different laboratories and in different occasions). This between-laboratory and between-assay CV was found to be 18.1% for IRMA Cis, 26.0% for IRMA Hybritech, and 26.9% for ICMA Immulite. The corresponding CVs observed in the same control samples for results of tPSA were markedly lower: 11.4% for IRMA Cis, 11.5% for IRMA Hybritech, and 17.2% for ICMA Immulite. The worse precision in measuring fPSA (with respect to tPSA) can be explained by the lower concentration of fPSA (on average 7–8% of tPSA in the control samples distributed in this survey) and suggests that fPSA methods are affected by scarce analytical sensitivity. This is confirmed by the behavior of precision in relation to fPSA concentration (precision profile). Samples with fPSA >0.5 µg/L show approximately constant CVs ranging from 14.9% to 19.1% (IRMA Hybritech), 12.3% to 16.4% (IRMA Cis), and 13.4% to 18.9% (ICMA Immulite). On the contrary, lower-concentration samples (<0.5 µg/L) exhibit precision that markedly worsened up to about 40% for all three methods.
To discriminate BPH from CAP patients, the fPSA determination is not
used alone, but combined with tPSA as the ratio fPSA/tPSA. For this
reason we evaluated the variability of the ratio fPSA/tPSA reported by
laboratories grouped by method. The CVs of the ratio have been computed
for two control pools (distributed in three occasions as hidden
replicates) with mean concentrations of tPSA of 4.25 and 8.94 µg/L
(see Table 1
). The variability of the ratio fPSA/tPSA was 21–36% in the lower pool
and 15–25% in the higher pool for the three methods considered. This
large variability is similar to that found for fPSA measurement; the
finding was expected since the CV of the ratio reflects both the CV of
the numerator and the CV of the denominator (according to the
well-known relation CVratio =
and the CV of fPSA is much larger (about twofold) than that of tPSA.
|
Moreover, the between-method differences of the ratio fPSA/tPSA (mean
values reported in Table 1
for the three methods) are larger than those
of fPSA alone. This is due to the fact that, for all three methods,
fPSA values do not appear directly correlated with the corresponding
tPSA values; for instance, IRMA Cis yelds the highest fPSA associated
with the lowest tPSA values. As a consequence the cutoff value of the
ratio fPSA/tPSA (used for clinical decision) has to be calculated in
each laboratory according to the methods used for fPSA and tPSA assay.
In conclusion, if the ratio fPSA/tPSA is to become a reliable tool in the clinical management of prostatic diseases, the precision of fPSA determination needs to be improved, particularly in the low range (<0.5 µg/L); in addition, a better standardization of different methods is desirable.
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  M. P. Fox, A. A. Reilly, and E. Schneider Effect of the Ratio of Free to Total Prostate-specific Antigen on Interassay Variability in Proficiency Test Samples Clin. Chem., August 1, 1999; 45(8): 1181 - 1189. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Gion, R. Mione, P. Barioli, M. Barichello, F. Zattoni, T. Prayer-Galetti, M. Plebani, G. Aimo, C. Terrone, F. Manferrari, et al. Percent free prostate-specific antigen in assessing the probability of prostate cancer under optimal analytical conditions Clin. Chem., December 1, 1998; 44(12): 2462 - 2470. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
) and ICMA Immulite (
) against results of the users
of IRMA Hybritech, for 22 control samples assayed during the 1996 cycle
of Oncocheck EQA.