| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Other Original Articles |
Department of Urology and Andrology (P.S.), Umeå University Hospital, 901 85 Umeå, Sweden; Department of Clinical Chemistry (U.-H.S.), Helsinki University Central Hospital, 00290 Helsinki, Finland; International Agency for Research on Cancer (E.R., R.K.), 69372 Lyon, France; and Department of Public Health and Clinical Medicine (G.H.), Umeå University Hospital, 901 85 Umeå, Sweden
Address all correspondence and requests for reprints to: Pär Stattin, Department of Urology and Andrology, Umeå University Hospital, 901 85 Umeå, Sweden. E-mail: par.stattin{at}urologi.umu.se
Abstract
Recent studies have suggested that IGF-I and IGF-binding protein (IGFBP)-3, in combination with prostate-specific antigen (PSA), may enhance prostate cancer detection. In this study, we sought to determine the effect on the prediction of future prostate cancer occurrence by incorporating ratios of total and free PSA, IGF-I, IGFBP-3 into PSA testing.
Within a population-based prospective cohort study, we investigated the validity (sensitivity and specificity) of plasma concentrations of total and free PSA, IGF-I, and IGFBP-3 and combinations thereof, in 114 cases and 97 controls, in the range of 1.75–13.5 µg/l for PSA, as used by Khosravi et al. (See Ref. 7 ).
Validity estimated by the area under the curve in receiver operator characteristics analysis (with 95% confidence interval) for total PSA was 0.78 (range, 0.71–0.84); total/free PSA, 0.69 (range, 0.62–0.76); total PSA/IGF-I, 0.72 (range, 0.65–0.79); free PSA/IGF-I, 0.55 (range, 0.48–0.63); total PSA/IGFBP-3, 0.74 (range, 0.68–0.81); and free PSA/IGFBP-3, 0.57 (range, 0.49–0.64).
Analysis of ratios of IGF-I, IGFBP-3, and free and total PSA did not improve validity of PSA testing in the prediction of future occurrence of prostate cancer. It is unlikely that these combinations will improve prostate cancer detection.
PROSTATE-SPECIFIC ANTIGEN (PSA) is the best circulating biomarker for cancer available today. However, there is a large overlap between prostate cancer and benign prostatic hyperplasia in patients with moderately increased levels of PSA (1, 2). For men with a normal finding on digital rectal examination and serum PSA between 4–10 µg/liter, there is a 20–30% risk of prostate cancer on sextant core biopsies, and a fairly similar risk has also been reported for men with PSA concentrations between 2–4 µg/liter (1, 3, 4). Consequently, much effort has been invested in increasing sensitivity and specificity of PSA testing. The mostly used enhancement of PSA testing is free PSA, and the ratio between free and total PSA has been shown to increase specificity by approximately 20–40% for PSA between 4–10 µg/liter (1, 4). To further improve prostate cancer detection, some authors have proposed that IGF-I should be added to PSA testing (5, 6, 7).
To test the hypothesis that ratios between IGF-I, IGF-binding protein (IGFBP)-3, and PSA can improve the sensitivity and specificity of PSA testing for prostate cancer, we analyzed these analytes for men with PSA in the range of 1.75–13.50 µg/liter, within a population-based prospective cohort study.
Materials and Methods
The design and methods used in this series have been described in detail (8). In brief, 29,660 men were recruited, between 1985–1999, to the Northern Sweden Health and Disease cohort study. By linkage with the Cancer Registry, we identified 149 incident cases of prostate cancer, who had donated plasma at recruitment (median time, 3.9 yr before diagnosis). Two controls per case were randomly selected from the cohort, matching the cases by age (±1 yr), and date (±1 yr) of recruitment.
Biochemical analyses
IGF-I and IGFBP-3 were measured by double-antibody, immunoradiometric assays (Immunotech, Marseille, France). The mean intraassay coefficients of variation were 13.5%, and 3.3%, respectively. Free and total PSA were determined by time-resolved immunofluorometric assays (Prostatus PSA; Wallac, Inc., Turku, Finland). The analytical detection limit of the assay was 0.01 µg/liter; and for values between 0.2 and 100 µg/liter, the inter- and intraassay coefficients of variation were 2–4%. We used the PSA range of 1.75–13.50 µg/liter, selected by Khosravi and colleagues (7) in the previous report on the same issue; 114 cases and 97 controls were within this range in our series. Serum levels of PSA shortly before the diagnosis were determined by the Tandem-R PSA assay (Hybritech, Inc., San Diego, CA) or the IMx PSA assay (Abbott Laboratories, Abbott Park, IL).
Statistical analyses
Pearson’s coefficients of correlation were used to examine the cross-sectional relationships between peptides. An unpaired t test was used to test for differences in mean levels of analyte concentrations between cases and controls. For analysis of intraindividual changes in PSA concentrations from the time of recruitment to diagnosis, a paired t test was used. Logistic regression analysis was used to calculate odds ratios (ORs) for prostate cancer by quartile levels of the peptides and combinations thereof. Cutpoints for quartiles were determined on variable distributions of cases and controls combined. To facilitate comparisons between ORs for combinations of analytes, the ratio that increased the ORs over increasing quartiles are presented. Using the SEs of pertinent regression coefficients, 95% confidence intervals (CIs) were computed. Validity (sensitivity and specificity) was estimated on the basis of area under the curve (AUC) calculated by receiver operator characteristics analysis. All statistical tests and corresponding P values were two-sided, and a P values less than 0.05 were considered statistically significant.
Results
Baseline characteristics
Cases and controls differed in several baseline characteristics
(Table 1
). Even in this restricted range
of PSA, mean plasma concentrations of PSA were significantly higher for
cases than for controls (5.55 and 3.27 µg/liter, respectively). Mean
values for ratios including total PSA differed significantly between
cases and controls, whereas no significant difference was detected for
means of the ratios free PSA/IGF and free PSA/IGFBP-3. For 106 of the
cases, PSA levels at the time of diagnosis were available; median time
from baseline to diagnosis was 4.4 yr (25th and 75th percentile,
2.3–6.1). Mean PSA concentrations increased from 5.43 µg/liter
(SD, 2.75) at baseline to 29.57 µg/liter (SD,
91.10) at the time of diagnosis, P < 0.008.
|
We combined data from cases and controls to examine the cross-sectional interrelationships between IGF-I, IGFBP-3, and free and total PSA. There were strong and significant correlations between total and free, total and free/total PSA, and free and free/total PSA (r = 0.64, -0.31, and 0.43, respectively). IGF-I and IGFBP-3 correlated strongly (r = 0.66), and both peptides correlated negatively to free/total PSA (r = -0.16). We did not find any significant correlation between IGF-I and IGFBP-3 to free or total PSA, nor did we find any correlations between IGF-I, IGFBP-3, and free and total form of PSA in separate analyses of cases and controls (data not shown). The PSA concentrations at baseline for cases did not correlate with PSA at the time of diagnosis.
Risk analysis
In logistic regression analysis, we found the strongest increase
in ORs for prostate cancer over increasing quartile concentrations of
total PSA, and all ratios involving total PSA were significantly
associated with risk (Table 2).
|
The validity in prostate cancer prediction estimated by the AUC
calculated by receiver operating characteristics curve analyses showed
that total PSA was the strongest predictor of future disease, in
comparison to all other analytes alone or in combinations. None of the
combinations had a stronger predictive value than PSA alone (Fig. 1 and Table 3). When analyzing all the
149 cases and 298 controls identified in
the original study (8), AUC (with 95% CI) was 0.91
(range, 0.88–0.94) for total PSA, 0.82 (range, 0.78–0.86) for free
PSA, 0.56 (range, 0.50–0.61) for IGF-I , and 0.54 (range, 0.49–0.60)
for IGFBP-3.
|
|
In this prospective cohort study, we found no improvement in prediction of future prostate cancer occurrence by the use of combinations of IGF-I, IGFBP-3, and free and total PSA, in comparison with a single measurement of total PSA.
Prediction of future prostate cancer diagnosis by PSA
PSA testing has dramatically changed the incidence and clinical management of prostate cancer (1). Most studies on PSA testing have related PSA levels to the presence of cancer in prostate biopsies performed shortly after the blood test (1, 2). In prospective studies such as ours, the ability of PSA to predict prostate cancer in subsequent years is investigated. In our study, as well as in other prospective studies (3, 9), a single measurement of PSA rather accurately predicted the occurrence of prostate cancer up to several years after testing, with an AUC of about 0.90. There was a 6-fold increase in mean PSA level from baseline to the time of diagnosis in our series, reflecting that blood levels of PSA mirror PSA production, as well as tumor development and progression, in the prostate. Consequently, it is very likely that PSA testing closer to the time of diagnosis would have had a higher validity in our series, as has been shown by others (10). Furthermore, the validity of PSA testing was reduced by the restriction in PSA range; in the full study group, the AUC was 0.91, compared with 0.78 in the restricted group.
IGF-I as a marker for prostate cancer
IGF-I has been associated with prostate cancer development in experimental studies (11), and elevated circulating levels of IGF-I has been associated with an increased risk of prostate cancer in three prospective studies (8, 12, 13). To improve the validity of PSA testing, IGF-I and IGFBPs have been investigated as biomarkers of prostate cancer in several recent cross-sectional studies (5, 6, 7, 14, 15). Notably, Khosravi and colleagues (7) investigated men with serum PSA between 1.75–13.50 µg/liter; 84 cases and 75 selected controls with benign prostatic hyperplasia, closely matched on total PSA, thereby minimizing the validity of total PSA. The ratios of IGF/free PSA and intact IGFB-3/free PSA marginally improved cancer detection, in comparison with free/total PSA; the AUC values were 0.73, 0.74, and 0.69, respectively. Djavan et al. (5) found that the ratio between IGF-I and total PSA increased prostate cancer detection, in comparison with PSA, in an analysis of 71 cases and 174 controls, with PSA between 2.5 and 15 µg/liter. AUC was 0.71 for IGF/PSA and 0.61 for PSA.
In contrast, Aprikian and colleagues (15), in their study of 546 men who underwent prostate biopsy because of elevated PSA concentrations or abnormal digital rectal examination results, found no significant difference in levels of IGF-I and IGFBP-3 between men with and without cancer. No predictive value was seen for IGF-I or IGFBP-3; AUC was 0.51 for IGF-I, 0.52 for IGFBP-3, 0.63 for IGF-I/PSA, 0.64 for PSA, and 0.76 for free/total PSA. Likewise, Finne and colleagues (14), in their analysis of 179 cases and 486 controls with PSA levels above 4 µg/liter detected in a population-based screening trial, found no increase in risk for elevated IGF-I levels, and AUC for IGF-I was 0.50. In that study, controls had, on average, larger prostates; and the authors suggested that IGF-I levels are related to prostate size rather than to the presence of cancer. In a prospective study, Harman et al. (12) found no value of IGF-I as a supplement to PSA for cancer detection in their study of 72 cases and 203 controls without restriction in PSA levels. Furthermore, they found no correlation between serum levels of IGF-I and prostate size.
Factors influencing the validity of tumor markers
Any estimate of sensitivity and specificity can be obtained by manipulating the selection of the target population (10). Obviously, the validity of the PSA test will decrease in analysis of men within a restricted range of PSA (5, 7); and by matching cases and controls closely on PSA concentrations, the validity of PSA will be drastically reduced (7). However, as shown by several studies (including ours), PSA continues to be a strong predictor of prostate cancer, even within rather restricted ranges (3, 16). For PSA, sensitivity is dependent on size, stage, and grade of the tumor, whereas specificity is largely dependent on the distribution of the size of the prostate. Prostate size increases with age, and men with voiding symptoms have, on average, a larger gland and will more likely seek medical advice and undergo prostate examinations. Furthermore, the predictive values of screening tests, which are crucial for the clinical utility of a test, depend on the prevalence of the disease in the study group, and they can be determined only in large, well-defined study groups, preferably with a distribution ratio of cases and controls and severity of disease reflecting that in the population at large.
IGF-I—tumor marker or etiological factor?
In a recent report, serum levels of PSA, IGF-I, IGFBP-2, and IGFBP-3 were serially measured after radical prostatectomy (17). In men with a relapse, PSA levels were initially high and continuously increased, whereas in men who remained free of disease, PSA levels remained low. However, the differences in IGF-I and IGFBP-3 levels were small between men who relapsed and men that did not, and no change over time occurred in concentrations in either group, showing that IGF-I and IGFBP-3 do not predict outcome after prostatectomy. In contrast, circulating levels of IGF-I have been moderately, but significantly, higher in cases than in controls in three prospective studies (8, 12, 13). Together, these studies suggest that circulating levels of IGF-I and IGFBP-3 do not reflect tumor production of IGF-I and IGFBP-3, and that IGF-I and IGFBP-3 are not markers of prostate cancer progression but possibly indicators of risk, as parts of an internal environment that may induce tumor development in the prostate.
Conclusions
Our data, in accordance with several other studies (12, 14, 15), strongly suggest that no increase in validity in prostate cancer detection is obtained by incorporating analysis of the IGF system to PSA testing. It is unlikely that analysis of IGF-I and IGFBP-3, in addition to PSA, will materially improve prostate cancer detection.
Acknowledgments
We thank all participants in the Västerbotten Intervention Project and Monitoring of Trends and Cardiovascular Disease Study.
Footnotes
This work was supported by grants from The Swedish Cancer Society; The Lions Research Foundation, Umeå; and The Medical Faculty, Umeå University, Umeå, Sweden.
Abbreviations: AUC, Area under the curve; CI, confidence interval; IGFBP, IGF-binding protein; OR, odds ratio; PSA, prostate-specific antigen.
Received May 23, 2001.
Accepted August 21, 2001.
References
This article has been cited by other articles:
 |
|
 |
|
  L. Peng, P. J. Malloy, and D. Feldman Identification of a Functional Vitamin D Response Element in the Human Insulin-Like Growth Factor Binding Protein-3 Promoter Mol. Endocrinol., May 1, 2004; 18(5): 1109 - 1119. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Colao, D. Ferone, P. Marzullo, and G. Lombardi Systemic Complications of Acromegaly: Epidemiology, Pathogenesis, and Management Endocr. Rev., February 1, 2004; 25(1): 102 - 152. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
