Stability of serum prostate-specific antigen determination across laboratory, assay, and storage time

ADULT

UROLOGY

STABILITY OF SERUM PROSTATE-SPECIFIC ANTIGEN DETERMINATION ACROSS LABORATORY, ASSAY, AND STORAGE TIME* STEVEN J. JACOBSEN, M.D., PH.D. GEORGE G. KLEE, M.D., PH.D. HANS LILJA, M.D., PH.D. GEORGE L. WRIGHT, JR., PH.D. JOSEPH E. OESTERLING, M.D. From the Department of Health Sciences Research, Section of Clinical Epidemiology, Department of Laboratory Medicine and Pathology, Division of Metabolic and Hematologic Biochemistry, Department of Urology, Mayo Clinic and Mayo Foundation, Rochester, Minnesota; the Department of Clinical Chemistry, University of Lund, Malmo, Sweden; and the Department of Microbiology and Immunology and Center for Urological Oncology, Eastern Virginia Medical School, Norfolk, Virginia

ABSTRACT-Objectives. To understand the comparability of serum prostate-specific antigen (PSA) determinations across assays and storage time. Methods. Serum PSA levels were determined for men aged 40 to 79 years from the clinical subset of the Olmsted County Study of Urinary Symptoms and Health Status Among Men on fresh samples and after a median of 32 months on banked samples, frozen at -70°C. Baseline serum PSA levels were determined by Tandem-R PSA assay. Follow-up levels on the banked samples were determined by the IMx PSA assay and a repeat Tandem-R PSA assay in a different laboratory and by an immunofluorometric PSA assay at another site. Results. The median serum PSA level determined by Tandem-R assay at baseline was 1 .O ng/mL (25th percentile, 0.6: 75th percentile, 1.7). The distributions of determination made by follow-up Tandem-R, IMx, and immunofluorometric analyses were essentially identical. Overall, the assays were highly correlated. The correlations between the baseline serum PSA determination and repeated Tandem-R, IMx, and immunofluorometric determinations were 0.96, 0.96, and 0.97, respectively (all P ~0.001). The median duration of frozen storage was 32 months (range, 26 to 39 months), and the correlations between baseline and follow-up determinations did not change when stratified by duration of storage. Conclusions. These data provide important reassurance about the use of serum PSA determinations obtained by different assays, in different laboratories, and in properly stored samples across time.

Prostate-specific antigen (PSA) is a serine protease produced by the epithelium of the prostate g1and.l Serum levels of PSA have been noted to be increased in persons with enlarged prostates2-4 and with *This work was supported in part by a grant from the Public Health Service, National Institutes ofHealth (AR30582), Merck Research Laboratories. and Abbott Laboratories. Submitted: August 22,‘1994, accepted: September 22, 1994

UROLOGY@ /MARCH~~~~IVOLUME~~,NUMBER~

prostate cancer. 2~5The former is presumably due to increased epithelial mass and the latter to tissue destruction and concomitant increased leakage into the systemic circulation.6 The measurement of serum levels has become extremely widespread, as serum PSA has proven most valuable in-monitoring response of prostate cancer to therapy7-‘r In addition, serum PSA determinations have been used increasingly for case finding and diagnosis.r2-l4 Some have

447

MATERIAL STUDY

Serum

1. Schematic

Status

1994

1989-1990 FIGURE

Sample

diagram

of study design.

even advocated the use of PSA for screening for prostate cancer,i5-l* but this view is not universally held. 19,20Despite this screening controversy, this use has received much attention. Public awareness has had a dramatic effect on clinical practice. Increasingly patients are requesting a serum PSA determination, and in light of guidelines from the American Cancer Society and the American Urological Association, physicians are ordering the tests. This can present the treating physician with a dilemma because of the number of different assays available and their poorly documented comparability Patients may present with PSA levels from different laboratories that may use different assays, Because small changes in serum PSA level (PSA velocity) may be interpreted as an indication for further diagnostic workup,21,22 it is especially critical to understand the comparability among assays and laboratories, especially at levels considered to be normal. Unfortunately, few data are available. Most published studies have been conducted in referral centers23-26 and have concentrated on cancer patients following treatment.26 None of the assays have been systematically compared in population-based cohorts. The stability of serum PSA in frozen serum is also poorly understood. Recently, investigators have analyzed stored serum from retrospectively identified cohorts.21 Although several reference laboratories have reported that levels are stable after 2 to 4 weeks 22,23,25there is no information available about stabiliiy over years. In this study, we report our findings of the comparability of three methods of measuring serum PSA concentration on samples from the baseline clinical cohort of The Ohnsted County Study of Urinary Symptoms and Health Status Among Men. Because these determinations were made in three different laboratories at different times using three different assays, we have the opportunity to determine if differences between assays and laboratories exist and to assess the stability of serum PSA over long-term storage.

448

AND METHODS

SUBJECTS

The design and implementation of this study have been described in detail elsewhere.3,27-29 In brief, between December 1989 and March 1991, the resources of the Rochester Epidemiology Project30 were used to identify an age- and residencestratified random sample of 5135 men aged 40 to 79 years from the Olmsted County, Minnesota, population. The medical records of these men were screened for a history of prostate cancer, prostatectomy, or specified conditions (other than benign prostatic hypertrophy [BPH]) that would interfere with voiding function, which eliminated 1261; the remaining 3874 (75%) were invited to participate in a prospective study of the natural history of prostatism. Of those identified, 2115 (55%) completed a previously validated questionnaire that assessed urinary symptoms, general health, and associated demographic characteristics, and voided into a standard urometer (Dantec 1000, Skovlunde, Denmark) to determine peak urinary flow rates. A random sample of 537 of these men (25%) was invited to the Mayo Clinic for a detailed urologic examination that included a serum PSA determination, meticulous digital rectal examination, and transrectal ultrasound determination of prostate volume. Of these, 475 men (88%) completed all three diagnostic tests. As a result of the evaluation, 52 (10%) underwent an ultrasound-guided biopsy of the prostate; 4 (7Oh) were found to have prostate cancer and were excluded from these analyses. PROSTATE-SPECIFIC ANTIGEN DETERMINATIONS

Prior to the clinical examination, study subjects presented at a central clinic facility for phlebotomy (Fig. 1). This procedure was performed in a nonfasting state and usually occurred in the morning hours. Approximately 15 mL of venous blood was drawn from each subject and divided into four aliquots of 3 to 4 mL each. Each aliquot was labeled and logged, and all but one was placed in frozen storage at -70°C. The unfrozen aliquot was processed immediately in a central laboratory to determine the serum PSA concentration with the Tandem-R PSA assay (Hybritech, San Diego, Calif). The results of this determination have been published previously3 In September 1993, one aliquot of serum was withdrawn from frozen storage for 395 (84%) of the 471 men. The frozen samples were transported on dry ice to Turku, Finland, where samples were thawed. The total serum PSA concentration was determined by an immunofluorometric assay

U ROLOGF

/ MARCH 1995 I VOLUME 45. NUMBER 3

10 9

IFMA

IMX

RTR

TandemR

A=Y FIGURE 2. Distribution

of serum PSA concentrations by assay. lmmunofluorometric assay (IFMA assay), IMx, and repeated Tandem-R (RTR) were performed on stored frozen serum samples with a median storage time of 32 months. Tandem-R (baseline Tandem-R determination) was performed on fresh serum at baseline. The box plot presents the median [hash-mark) 1 st and 3rd quartile (limits of box) and 5 % and 95 % percentiles [limits of whiskers).

(IFMA)31,32; the remaining serum was refrozen and transported on dry ice to Eastern Virginia Medical School where the serum PSA concentration was determined using both the Tandem-R and IMx (Abbott Laboratories, North Chicago, Ill) PSA assays.33

STATISTICALANALYSES The primary measure of association for analyses was a Pearson product moment correlation coefficient. This measure can range from -1 (perfect

TABLE I.

inverse association) to +l (perfect direct association) with 0 representing no linear association, Because of the log normal distribution of PSA levels, the natural logarithm of the PSA level was used for analyses to meet the assumption of bivariate normality To provide further insight, the analyses were repeated, stratified by serum PSA level and length of time in frozen storage. Because it has been suggested that the correlation coefficient is an inadequate measure of agreement,34 serum PSA concentrations were graphically compared in a pairwise fashion, plotting the difference in measurement on the ordinate and mean of the two measures in the abscissa.35Paired t tests were calculated to test for bias (systematic differences) with the Bonferroni correction for multiple comparisons. 36 The correlation coefficient was then calculated for each of these pairwise comparisons as a test of consistency of bias.37 The slope and intercept describing each of the pairwise comparisons of serum PSA determinations were estimated using ordinary least-squares regression models. Ail analyses were performed using SAS statistical software (Gary, NC). RESULTS In comparing the 395 subjects for whom repeat analyses were available with the 76 for whom repeat analyses were not, there were no differences in the distribution of ages or PSA concentrations. The median (first, third quartile) duration of frozen storage was 32 months, with a minimum of 26 and maximum of 39 months. The distribution of each of the PSA determinations by the various assays is presented in Figure 2. The median serum PSA concentration by the Tandem-R PSA assay at baseline was 1.0 ng/mL with the 25th and 75th percentiles being 0.6 and 1.7 ng/mL, respectively The distribution of IFMA-determined values was similar (median, 0.9 ng/mL; Ql, 0.6 ng/mL; Q3, 1.5 ng/mL),

Correlation among serum PSA determinations on same sample

Assay lmmunofluorometric IMx Repeat Tandem-R

Baseline Tandem-R r n r n r n

0.97 395 0.96

Assay lmmunofluorometric Overall

IMx

0.99

384

384

0.96

0.9%

0.97

384 388 r = Pearsonproduct-mommtuwrclation wefficie~~ts drlctrnlnedftomfhr IIULUI.UI logar-ithmofthr sewn PSAlevel n = numberofpail-swith Iwrls uvailublcforanalyGs All correlationcoefficwtswew slutisticallysqn$unlt, P CO001

UROLOGY@ / MARCH 1995 I VOLU~M~45, NUMBER3

388

449

TABLE II. determination

Correlation of subsequent on same sample stratified

PSA assays with baseline by length of frozen storage Storage 30-3 1

26-29 lmmunofluorometric

assay

r n r n r n

IMx Tandem-R

0.98 93 0.98 90 0.97 92

0.98 101 0.96 99 0.96 99

r = Pearson product-moment correlation coejjicients determined from n = number ojpairs with levels available for analysis All correlation coefficients were statistically si@ficant, P
as was the distribution for the IMx-determined levels (median, 1.0 ng/mL; Ql, 0.7 ng/mL; 43, 1.7 ng/mL) and the concentrations determined by the repeat Tandem-R PSA assay (median, 0.9 ng/mL; Ql, 0.5 ng/mL; 43, 1.6 ng/mL). Overall, the results obtained with the various assays were similar, with correlation coefficients ranging from 0.96 to 0.99 (Table I, all P
TABLE III.

Correlation

Baseline

n r n

Tandem-R 1.1-2.0

Tandem-R

Serum PSA concentration lmmunofluorometric

2.1-3.0

IMx n Tandem-R ri Serum PSA concentration lmmunofluorometric

> 3.0

r n n

Tandem-R n

r = Pearson product-moment correlation coefficients determined from n = number ojpairs with levels availablefor analysis. All correlation coefficients were statistically signijicant. P < 0.001.

450

PSA level.

onsame

sample,

by

Assay lmmunofluorometric Assay

Tandem-R

IMx

0.88 192 0.85 187 0.82 190

0.94 187 0.91 190

0.86 187

0.73 120 0.68 116 0.7 1 116

0.93 116 0.89 116

0.86 116

0.90 39 0.71 38 0.74 38

0.83 38 0.78 38

0.66 38

0.97 44 0.96 43 0.91 44

0.95 43 0.91 ‘I 4

0.88 44

ng/mL

IMx Repeat

serum

ng/mL n

Repeat

ojthe

ng/mL

IMx Repeat

logarithm

0.96 99 0.95 96 0.95 96

0- 1 .O ng/mL

IMx

Serum PSA concentration Irnmur~ofluorometric

35-39

When follow-up assays were stratified by length of storage time, the correlations with the original Tandem-R PSA assay were essentially unchanged (Table II). For example, the correlation between the follow-up and the original Tandem-R PSA assays was 0.97 for samples stored for 2 years and 0.95 for samples stored over 3 years. Furthermore, when stratified by baseline PSA levels (Table III),

n

Repeat

the natural

(mos] 32-34 0.98 102 0.98 99 0.98 101

among serum PSA determinations baseline concentration

Assay Serum PSA concentration lmmunofluorometric

Time

the natural

logarithm

of the sennn

UROLOGY@

PSA kwl

/ MARCH

1995

i VOLUME

45,

NUMBER

3

TABLE IV.

Pairwise comparisons Mean Difference+ 0.165 -0.03” 0.109

of serum prostate-specific 2.5th. 97.5th* Percentiles -0.13, 0.67 -0.63, 0.44 -0.64, 0.66

Baseline Tandem-R-IFMA Baseline Tandem-R-IMx Baseline Tandem-Rfollow-up Tandem-R IFMA-IMx -0.195 -0.86, 0.08 IFMA-follow-up Tandem-R -0.055 --1.04, 0.29 0.144 -0.56, 0.91 IMx-follow-up Tandem-R *Resultsof leastsquaresregressionof secondmonber ofpair onfirst monber ofpair. ‘Mean ofpairwise diJ@ferences. +Percentiles ofpainvisedifiereerences. “P ~0.01,paired t testwith Bonferronicorrection. I’P~0.05,paired t testwith Bonferronicorrection. KEY:SE= standard~TTOI;RMSE= motmeansquareenvr aboutregressionline.

antigen

determinations

Slope 0.93 0.91 1.03

SE 0.01 0.01 0.02

Regression Analyses Intercept -0.09 0.06 -0.12

* SE 0.01 0.01 0.01

RMSE 0.17 0.20 0.24

0.97 1.10 0.85

0.01 0.01 0.01

0.15 -0.02 0.16

0.01 0.01 0.01

0.1 1 0.18 0.19

the correlation remained high at the higher baseline concentration despite the small numbers. The pairwise comparison of differences (Table IV) demonstrates a striking comparability among the assays. The mean difference between any two assays was less than 0.2 ng/mL, and all but one of the 2Sth or 97Sth percentiles of the difference were less than 1.0 ng/mL from zero (Fig. 3). Serum levels determined by the IFMA assay were consistently lower than the other assays (P ~0.01); the correlations suggested that this was more pronounced at high PSA levels. Regression models of serum PSA on age suggested that the nominal age-specific cutpoints were similar across all the assays except the IFMA, which was slightly lower at all ages (2.0 versus 2.5 for men aged 40 to 49 years; 3.0 versus 3.5 for those aged 50 to 59 years; 4.0 versus 4.5 for those aged 60 to 69 years; and 5.5 versus 6.5 for those aged 70 to 79 years). COMMENT

-2-

I

I

I

I

123456 FIGURE 3. Distribution of pairwise differences between serum PSA determinations. Pair 1: TR (baseline Tandem-R) and IFMA (immunofluorometric assay); Pair 2: TR and IMx; Pair 3: TR and RTR (repeated TandemR); Pair 4: IFMA and IMx; Pair 5: IFMA and RTR; Pair 6: IMx and RTR. IMx and RTR were performed on stored frozen serum samples with a median storage time of 32 months. Tandem-R was performed on fresh serum at baseline. The box plot presents the median (hash-mark) 1st and 3rd quartile [limits of box) and 5% and 95% percentiles (limits of whiskers).

U ROLOGYa I MARCH 1995 I VOLUME 45, NUMBER3

These data provide important reassurance about serum PSA determinations in light of the increased use in clinical care and research. Serum PSA determinations appear to be consistent across the commercially available assays and appear stable over long-term frozen storage. These results have several important implications. First, the finding that serum PSA levels were measured consistently with the IMx or Tandem-R PSA assays suggests that comparisons across assays, whether for patient care or research, are valid. Furthermore, these results indicate that PSA values determined in different laboratories that use standard quality control techniques and the commercially available assays are similar and comparable. The second major implication from this study is that the measurement of serum PSA concentrations appears to be 451

stable and unbiased in serum frozen at -70°C for more than 3% years. This could prove especially important for studies that have stored serum samples for future analyses. 23 This opens the door for additional retrospective cohort studies or nested case-control studies of PSA to be undertaken. Although these data offer many reassurances, there are several important limitations that should be kept in mind. This study did not use a full factorial design; not all possible combinations of assay and storage time were tested. Only one laboratory used the IMx PSA assay and a different laboratory utilized the IFMA. Two different laboratories used the Tandem-R PSA assay, one on fresh serum and the other on serum frozen at -70°C. Thus, the results obtained could have occurred if a systematic bias in measurement existed in opposing directions and canceled each other. Although possible, this seems unlikely. Another potential limitation stems from the fact that these results were obtained from a community-based sample of white men. Generalization to other races and cultural settings may not be appropriate. Nonetheless, these data from a community-based cohort (neither from referral practices nor strictly cancer patients) provide a baseline comparison that others can attempt to replicate or refute, both in similar populations or others. Steven J. Jacobsen, M.D., Ph.D. Department ofHealth Sciences Research Mayo Clinic 200 First Street 5. W. Rochester, MN 55905 ACKNOWLEDGMENT. To Sondra Buehler and Yvonne Weeldreyer for their help in manuscript preparation; to Drs. Harry A. Guess, Christopher G. Chute, and L. Joseph Melton, III, for their contributions to the design of this study, and to Michelle Morningstar and Mary Ann Clements for perforning the repeat Tandem-R and IMx assays.

REFERENCES I. Papsidero LD, Kuriyama M, Want MC, Horoszewicz I, Leong SS, Valenzuela L, Murphy GP, and Chu TM: Prostate antigen: a marker for human prostate epithelial cells. J Nat1 Cancer Inst 66: 3742, 1981. 2. Stamey TA, Yang N, Hay AR, McNeal JE, Freiha FS, and Redwine E: Prostate-specific antigen as a serum marker for adenocarcinoma of the prostate. N Engl J Med 317: 909-916, 1987. 3. Oesterling JE, Jacobsen SJ, Chute CG, Guess HA, Girman CJ, Panser LA, and Lieber MM: Serum prostate-specific antigen in a community-based population of healthy men: establishment of age-specific reference ranges. JAMA 270: 860-864, 1993. 4. Collins GN, Lee RJ, McKelvie GB, Rogers AC, and Hehir M: Relationship between prostate specific antigen,

452

prostate volume and age in the benign prostate. Br J Urol 71: 445450, 1993. 5. Oesterling JE: Prostate specific antigen: a critical assessment of the most useful tumor marker for adenocarcinoma of the prostate. J Urol 145: 907-923, 1991. 6. Brawer MK, and Lange PH: Prostate specific antigen: its role in early detection, staging, and monitoring of prostatic carcinoma. J Endourol 3: 227-236, 1989. 7. Landmann C, and Hunig R: Prostate specific antigen as an indicator of response to radiotherapy in prostate cancer. Int J Radiat Oncol Biol Phys 17: 1073-1076, 1989. 8. Stamey TA, Kabalin JN, McNeil JE, Johnstone IM, Frieha F? Redwine EA, and Yang N: Prostate specific antigen in the diagnosis and treatment of adenocarcinoma of the prostate. II. Radical prostatectomy treated patients. J Urol 141: 1076-1083, 1989. 9. Lange PH, Ercole CJ, Lightner DJ, Fraley EE, and Vessella R: The value of serum prostate specific antigen determinations before and after radical prostatectomy. J Urol 141: 873-879, 1989. 10. Stamey TA, Ferrari MK, and Schmid H-P: The value of serial prostate specific antigen determinations 5 years after radiotherapy: steeply increasing values characterize 80% of patients. J Urol 150: 1856-1859, 1993. 11. Ruckle HC, Klee GG, and Oesterling JE: Prostatespecific antigen: concepts for staging prostate cancer and monitoring response to therapy. Mayo Clin Proc 69: 69-79, 1994. 12. Cooner WH, Mosley BR, Rutherford CL Jr, Beard JH, Pond HS, Terry WJ, Igel TC, and Kidd DD: Prostate cancer detection in a clinical urological practice by ultrasonography, digital rectal examination and prostate specific antigen. J Urol 143: 1146-1154, 1990. 13. Ellis W, and Brawer M: The role of tumor markers in the diagnosis and treatment of prostate cancer, in Lepor H, Lawson RK (Eds): Prostate Diseases. Philadelphia, WB Saunders, 1993, pp 276-292. 14. Rommel FM, Agusta VE, Breslin JA, Huffnagle HW, Pohl CE, Sieber PR, and Stahl CA: The use of prostate specific antigen and prostate specific antigen density in the diagnosis of prostate cancer in a community based urology practice. J Urol 151: 88-93, 1994. 15. Mettlin C, Jones G, Averette H, Gusberg SB, and Murphy GP: Defining and updating the American Cancer Society guidelines for the cancer-related checkup: prostate and endometrial cancers. CA Cancer J Clin 43: 42-46, 1993. 16. American Urological Association: Early detection of prostate cancer and use of transrectal ultrasound, in American Urological Association 1992 Policy Statement Book. Baltimore, American Urological Association, 1992, p 4.20. 17. Catalona WJ, Smith DS, Ratliff TL, and Basler JW: Detection of organ-confined prostate cancer is increased through prostate-specific antigen-based screening. JAMA 270: 948-954, 1993. 18. Catalona WJ: Screening for prostate cancer: enthusiasm. Urology 42: 113-115, 1994. 19. Robbins AS: PSA and the detection of prostate cancer. JAMA 271: 192-193, 1994. 20. Chodak GW: Questioning the value of screening for prostate cancer in asymptomatic men. Urology 42: 116-118, 1994. 21. Carter HB, Pearson JD, Metter EJ, Brant LJ, Chan DW, Andres R, Fozard JL, and Walsh PC: Longitudinal evaluation of prostate-specific antigen levels in men with and without prostate disease. JAMA 267: 2215-2220, 1992. 22. Brawer MK, Beatie J, Wener MH, Vessella RL, Preston SD, and Lange PH: Screening for prostatic carcinoma with

UROLOGY@ /MARCH

1995 I VOLUME 45. NUMBER 3

prostate specific antigen: results of the second year. J Urol 150:106-109,1993. 23. Klee GG, Dodge LA, Zincke H, and Oesterling JE: Measurement of serum prostate-specific antigen concentration using the IMx PSA assay. J Urol 151: 94-98, 1994. 24. Dnistrian AM, Schwartz MK, Smith CA, Nisselbaum JS, and Fair WR: Abbott IMx~ evaluated for assay of prostatespecific antigen in serum. Clin Chem 38: 2140-2142, 1992. 25. Vessella RL, Noteboom J, and Lange PH: Evaluation of the Abbott IMx@ automated immunoassay of prostate specific antigen. Clin Chem 38: 2044-2054, 1992. 26. Kuriyama M, Akimoto S, Akaza H, Arai Y-I, Usami M, Imai K-I, Tanaka Y, Yamazaki H, Kawada Y, Koiso K, et al: Comparison of various assay systems for prostate-specific antigen standardization. Jpn J Clin Oncol22: 393-399, 1992. 27. Jacobsen SJ, Guess HA, Panser LA, Girman CJ, Chute CG, Oesterling JE, and Lieber MM: A population-based study of health care-seeking behavior for treatment of urinary symptoms: the Olmsted County Study of Urinary Symptoms and Health Status Among Men. Arch Fam Med 2: 729-735, 1993. 28. Chute CG, Panser LA, Girman CJ, Oesterling JE, Guess HA, Jacobsen SJ, and Lieber MM: The prevalence of prostatism: a population-based survey of urinary symptoms. J Urol 150:85-89,1993. 29. Panser LA, Chute CG, Girman CJ, Guess HA, Oesterling JE, Lieber MM, and Jacobsen SJ: Effect of several recruitment strategies on response rates at baseline in a prospective cohort investigation: the Olmsted County Study of Urinary Symptoms and Health Status Among Men. Ann

U ROLOCY@ / MARCH 1995 I VOLUME 45, NUMBER 3

Epidemiol 4: 321-326, 1994. 30. Kurland LT, and Molgaard CA: The patient record in epidemiology. Sci Am 245: 54-63, 1981. 31. Lilja H, Christensson A, Dahlen U, Matikainen M-T, Nilsson 0, Pettersson K, and Lovgren T: Prostate-specific antigen in serum occurs predominately in complex with orantichymotrypsin. Clin Chem 37: 1618-1625, 1991. 32. Lilja H, Bjork T, Abrahamsson P-A, Stenman UH, Shaw N, Dowel1 B, Oesterling JE, Pettersson K, Piironen T, and Ldvgren T: Improved separation between normals, benign prostatic hyperplasia (BPH), and carcinoma of the prostate (CAP) by measuring free (F), complexed (C) and total concentrations (T) of prostate specific antigen. J Urol 151: 400A, 1994. 33. Fiore M, Mitchell J, Doan T, Nelson R, Winter G, Grandone C, Zeng K, Haraden R, Smith J, Harris K, et al: The Abbott IMx@ automated benchtop immunochemistry analyzer system. Clin Chem 34: 1726-1732, 1988. 34. Altman DG, and Bland JM: Measurement in medicine: the analysis of method comparison studies. Statistician 32: 307-317, 1983. 35. Bland JM, and Altman DG: Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1: 307-310, 1986. 36. Fleiss JL: The Design and Analysis of Clinical Experiments. New York: John Wiley 6: Sons, 1986, p 233. 37. Oesterling JE, Chute CG, Jacobsen SJ, Guess HA, Panser LA, Johnson CL, and Lieber MM: Longitudinal changes in serum PSA (PSA velocity) in a community-based cohort of men. J Urol 149(suppl): 412A, 1993.

453