STABILITY OF SERUM TOTAL AND FREE PROSTATE SPECIFIC ANTIGEN UNDER VARYING STORAGE INTERVALS AND TEMPERATURES

0022-5347/97/1586-2 182$03.00/0 THE JOURNAL OF

Vol. 158.2182-2187. December 1997 Printed rn U.S.A.

uROi.OCY

Copyright 8 1997 by AWEKIVAV U K O L ~ X : AWM-IATIOS, ICAI. 1st

STABILITY OF SERUM TOTAL AND FREE PROSTATE SPECIFIC ANTIGEN UNDER VARYING STORAGE INTERVALS AND TEMPERATURES CARLOS G. ARCANGELI,* DEBORAH S. SMITH, TIMOTHY L. RATLIFF WILLIAM J. CATALONA

AND

Froin the Division of Urologic Surgery, Department of Surgery. Washington Uniuersitv School of Medicine, St Louis, Missouri. and Departnient of Urology, U~iiversrtyof Iowa College of Medicine, louia City, Iowa

ABSTRACT

Purpose: Measurement of total serum prostate specific antigen (PSA) is widely used as an aid to early detection of prostate cancer. Measurement of the ratio of free-to-total PSA (percentage of free PSA) may help increase specificity of PSA testing. We prospectively studied the effects of varying the storage temperature and interval on total and free PSA levels. Materials and Methods: We measured the baseline total and free serum PSA levels in 36 volunteers (mean age 66 years) and then retested aliquots of these serum samples after varying storage intervals (24 hours, 2 weeks and 9 months) a t 3 different temperatures (4C, -2OC and -7OC). Volunteers represented a spectrum of prostatic conditions (PSA levels 2.0 to 4.0 ng./ml., PSA levels greater than 4.0 ng./ml. without cancer and PSA levels greater than 2.0 ng./ml. with prostate cancer). We used repeated measures analysis of variance to test for changes in total and free PSA levels as a function of time and temperature. We also evaluated the impact of storage a t different temperatures and times on the percentage of free PSA. Results: Across groups total and free serum PSA decreased from the baseline level differentially as a function of longer storage interval and higher temperature ( p (0.05). No significant difference was found for change in total PSA a t 24 hours, 2 weeks or 9 months for storage temperatures of -20C compared with -70C. A significant change from baseline level was found for free PSA when stored in -2OC compared with -7OC for 2 weeks but the magnitude of the change was modest. Conclusions: For storage intervals up to 9 months total PSA is more stable than free PSA under temperature conditions ranging from 4C to - 70C. This differential stability has important implications for the clinical evaluation of percentage of free PSA to distinguish between benign and malignant diseases of the prostate. KEY WORDS:prostatic neoplasms, prostate-specific antigen

Prostate cancer is the most common noncutaneous malig- serum total and free PSA levels.10.lI Ejaculation also may nancy in American men.' The widespread use of serum prostate increase transiently total and free PSA levels.I2 Additionally, specific antigen (PSA) as an aid for early cancer detection has fluctuations in total and free PSA serum levels occur within accounted for much of this increase in prostate cancer preva- individuals tested repeatedly during short intervals, with lence.2 PSA based screening detects prostate cancer at an ear- variability estimates ranging from 7 to 58o/c.l".1 1 . * 3 lier stage,:<..'increasing the likelihood that curative therapy can Other factors that may affect serum PSA levels are the be offered. To detect cancer while it is still potentially curable, handling procedures and conditions under which serum is the PSA cutoff used for recommendation of prostatic biopsy stored. Previous investigators have found that free PSA is must be low. However, low PSA cutoffs produce an appreciable less stable than total PSA under long-term storage condiproportion of false-positive PSA test results. t i o n ~ .14.~ , However, these studies are limited by either PSA exists in several different forms in serum. Most PSA small sample sizes or the lack of clearly defined patient in serum is bound to proteins and only a small fraction is groups. The purpose of this report is to evaluate in well unbound (free)."" Men with prostate cancer have a lower characterized patient groups the change in total, free and percentage of free PSA in serum than those with nonmalig- percentage of free PSA serum levels after storage for 24 nant conditions of the p r ~ s t a t e 5). ~Therefore, measurement hours, 2 weeks and 9 months under 3 temperature conditions of the percentage of free PSA in serum increases the speci- (4C. -2OC and -7OC). ficity of PSA screening. To optimize accuracy in interpretation of serum total and MATERIALS A N D METHODS free PSA levels for clinical and research purposes, factors that may spuriously increase or decrease test results have Subjects and procedures. We obtained serum samples from been examined. For example, prostatic manipulation, such as men enrolled in a prostate cancer screening study and from digital rectal examination and prostatic biopsy, can affect patients referred with the diagnosis of prostate cancer. As described previously,'" our screening volunteers undergo digAccepted for publication May 29, 1997. ital rectal examination and PSA testing at 6-month intervals. Supported in part by a grant from Iivbritcch, Inc.. San Diego. We recommend transrectal ultrasound guided biopsy for Californin. Rec u e s t i for reprints: Division of Urologic Surgery. 4960 <'hil- those with suspicious findings on either test. Between Sepdrrn's b l a St. ~ ~Ifiuis. ~ Missouri 69110. tember and October 1995 we recruited 59 men who presented +

2182

2183

STABILITY OF SERUM TOTAL AND FREE PROSTATE SPECIFIC ANTIGEN for regularly scheduled screening tests, including 40 men with at least 2 previous total serum PSA levels 4.0 or less ng./ml., and 19 men with at least 2 previous total serum PSA levels greater t h a n 4.0 and a t least 2 prostatic biopsies t h a t were negative for prostate cancer. The cancer study group included 13 men with a diagnosis of prostate cancer who were scheduled to undergo radical prostatectomy at our institution during the same interval. Since the reproducibility of low free PSA levels was of concern, we excluded from further analysis 31 men for whom the total PSA level was less than 2.0 ng./ml. based on results from t h e fresh serum sample. Therefore t h e study group included 11 men with PSA levels between 2.0 and 4.0 ng./ml., 19 with PSA levels greater than 4.0 ng./ml. without prostate cancer and 11 with PSA levels greater t h a n 2.0 ng./ml. with cancer. The table presents the mean age at the time of the blood draw and initial median total and free PSA stratified by study group. The mean age of the men with prostate cancer was less (62 years) t h a n t h a t of the other study groups (mean age 67 and 69 years for men with PSA 2.0 to 4.0 ng./ml. and men with PSA greater t h a n 4.0 ng./ml. without prostate cancer, respectively, p = 0.06).As per the study design t h e median initial PSA level was significantly lower in the men with PSA levels between 2.0 and 4.0 ng./ml. (3.0 versus 5.1 ng./ml. for men with PSA greater t h a n 4.0 ng./ml. without cancer and 5.4 ng./ml. for men with prostate cancer, p = 0.0002). A similar pattern also was found for median free PSA level ( p = 0.03). Of the 41 volunteers 3 were not white (7%). We obtained approximately 60 ml. blood from each study volunteer. The blood was transported to a central laboratory and processed within 3 hours. Processing the blood required centrifugation a n d separation of t h e sera for division into 1 ml. aliquot tubes. One free and 1 total PSA measurements were performed immediately, and all results on stored samples were compared to these initial values. A third of each volunteer serum was stored at each of the 3 temperatures (4C, -2OC and -7OC) for future free and total PSA measurements. We collected sufficient serum from each volunteer to perform free and total PSA measurements at each temperature for 3 subsequent time periods of 24 hours, 2 weeks and 9 months, for a total of 9 time x temperature combinations (fig. 1). Since t h e sera were aliquoted into small vials, there was no need to thaw t h e entire batch of serum for each measurement. Therefore, our samples were thawed only at the time of t h e assay. Total arid free assay. We measured total serum PSA concentrations using a n immunoenzymetric assay (Tandem-E" PSA).:jWe measured serum free PSA concentration using the Tandem-R* free PSA assay.8 All samples were run in duplicate for the total and free PSA assay, with the mean of the replicates reported a s t h e final result. Samples with a coefficient of variation among replicates greater than 10%were considered invalid. For each volunteer a total of 20 assays was completed ( 10 for total PSA and 10 for free PSA, t h a t is, 1 initial test and 9 time >: temperature combinations). Of the 41 men 5 had 3 or more assay results (of 20 total tests) t h a t were considered invalid because of high coefficient of variation among repli' Hybritech, Inc.. San Diego, California.

I

'

I

allquo1 rubes.

I

I

PSt\ mcasured / ~ j

Free and Total

PS 9 merisured 'd

I r w and I oral

PSA measured ti? 9 months

FK. 1. Study protocol for measuring serum total and free PSA at varying storage intervals and temperatures. cates in the free PSA assay. The 5 men were excluded from the final study group. The excluded men did not differ from t h e remainder of t h e volunteers with regard to age, initial total PSA level or initial free PSA level (Mann-Whitney U p >0.3). An additional 6 men had 1 free PSA assay result t h a t was considered invalid. Since t h e invalid result represented only 1 of 10 of the free assays performed for each man and since we did not want to eliminate these 6 cases from t h e overall analyses, we substituted for the invalid assay result free PSA values based on t h e mean valuesr7 found for each storage interval and temperature condition within the individual respective study group. None of t h e invalid results was from the same storage interval a n d temperature condition. Overall, 0.2% of the total PSA results tl of 4101were considered invalid and 5.4%of the free F'SA tests ( 2 2 of 410) were invalid. Percentage of free PSA was calculated as the ratio of freeto-total PSA multiplied by 100. Percent change for total and free PSA was calculated a s the difference between the initial serum level and the stored level, divided by the initial level.

Mean age and niediari total a d free PSA hasid on f r m h serum saniplr stratified h.v stlrdv group Proatate Ca PSX Greater Than 4 Ng.AII PSA 2 to 4 Ng./hll (11 DtS.1 nlean age 1 r S D I

Median total PSA 1 :SIR)? Median frpp PSA 1 +SIRIT ~

'

~~~

66.7 I t6.61 3.0 ( 2 0 . 3 ) 0.7 I10.31

Represent %way comparisons via Kruskal-Wallis tests. Semi-interquartile range (175th percentile25th percentilel/2).

I

I19 pta.1

fiH.5I r 7 . 3 1 5.1 110 .5l 0.9 ~ 0 . 3 1

I 11

pts

I

62.1 I 16.81 5.4 I t2.71 1.6 1 ~ 0 . 3 1

p villue' 0.06 0.0002 0.03

2184

STABILITY O F SERUM TOTAL AND FREE PROSTATE SPECIFIC ANTIGEN

Statistical analysis. We plotted median percent change for median percent change in total PSA after storage of 2 weeks total, free and percentage of free PSA from initial serum or 9 months did not differ a s a function of whether the serum levels for the 9 storage conditions. We originally computed a was stored in -2OC o r -7OC conditions. Median percent change in free PSA a s a function of storage 3 (study group) x 3 (storage interval) x 3 (storage temperature) repeated measures analysis of variance to test for dif- interval and temperature. The median percent change in ferences in percent change with time for total and free serum serum free PSA showed a decrease from initial levels in all PSA at the 3 storage temperatures and to determine whether storage temperatures after the 24-hour storage interval the changes with time at the different storage temperatures (overall median percent change ranged from - 5 to -9% a t 24 differed by study group. Since we did not find a significant hours, fig. 3 ) . Similar to the results for total PSA, after a time X temperature X study group interaction for either storage interval of 2 weeks or 9 months, the median percent total or free PSA ( p values for interactions >0.2), we simpli- change was markedly different for serum stored in 4C (mefied the analysis by recomputing the repeated measures dian percent change - 2 8 9 for 2 weeks and -50% for 9 analysis of variance, including only the within-group factors months) compared with those stored in temperatures of of interval and temperature. To investigate further signifi- -2OC (median percent change -8% for 2 weeks and -8% for cant interval x temperature 2-way interactions, we used 9 months) or -7OC (median percent change -5% for 2 weeks post-hoc comparisons to test for significant differences in and -9% for 9 months). In contrast to total PSA, the serum percent change for specific storage interval and temperature free PSA level decreased to a greater extent after 2 weeks of combinations selected a priori based on clinical relevance. storage at -2OC compared with 2 weeks of storage at -7OC. However, the range of percent change was greater at -2OC RESULTS (range 1 to -50%) compared with -7OC (range 1to -2O%), Median percent change in total PSA as a function of storage indicating substantial variability in percent change a t -2OC interval and temperature. The median percent change in a t the 2-week time point. The median percent change at 9 total PSA after 24 hours of storage was minimal and did not months was similar with the 2 temperature conditions. Similar to the results for total PSA, the repeated measures differ as a function of storage temperature (fig. 2). However, for a storage interval of 2 weeks or 9 months the median analysis of variance for free PSA showed a significant 2-way percent change was markedly different for serum stored at interaction for storage interval and temperature (F[4,140) = 4C (median percent change -7% for 2 weeks and -20% for 9 108.4, p 0.5). This result indicated that the PSA as a function of storage conditions. After storage of only

0

14

270

Time (Days) Fic:. 2. hledian percent decrease in serum total PSA in 36 study volunteers as function of storage interval and temperature. Error bars correspond to semi-interquartile range ( 175th percentile-25th percentileV2) for each interval and temperature condition.

STABILITY OF SERUM TOTAL AND FREE PROSTATE SPECIFIC ANTIGEN

2185

FIG.3. Median percent decrease in serum free PSA in 36 study volunteers as function of storage interval and temperature. Error bars correspond to semi-interquartile range ( [75th percentile-25th percentilel/2) for each interval and temperature condition.

24 hours i n any of t h e temperature conditions, the percentage of free PSA level may decrease 4 to 9% from the value that would have been found with fresh serum (fig. 4). This decrease in percentage of free PSA is a function of the fluctuation in total PSA after storage of 24 hours and the median negative percent change in free PSA after storage of 24 hours (figs. 2 a n d 3). After storage of 2 weeks there is a greater decrease in percentage of free PSA with conditions of -20C versus -7OC (median percent change -10% a t -2OC a n d -3% a t -7OC). This results from t h e proportionally greater decrease i n free PSA versus total PSA in storage conditions of -2OC compared with -7OC (figs. 2 and 3). Therefore, al-

though total PSA does not differentially decrease a t -2OC compared with -7OC, t h e differential decrease in free PSA at t h e higher temperature can affect the percentage of free PSA. After 9 months of storage t h e change in percentage of free PSA is similar for storage temperatures of -20C or -70C (fig. 4). DISCUSSION

The percentage of serum PSA that exists in the free form is lower i n men with prostate cancer than in those with benign prostatic hyperplasia. This disparity can be exploited clini-

FIG. 4. Median percent decrease in serum percentage of free PSA in 36 study volunteers as function of storage interval and temperature. Error bars correspond to semi-interquartile range ([75th percentile-25th percentileV2) for each interval and temperature condltlon. FIT, free-to-total.

2186

STABILITY OF SERUM TOTAL AND FREE PROSTATE SPECIFIC ANTIGEN

cally to distinguish between prostate cancer and benign dis- 4CT, resulting in proportionately greater levels of measeases of the prostate."-9 However, many reports on free PSA urable total PSA compared with free PSA. Alternatively, as have been retrospective and based on frozen serum that has suggested by Woodrum et al,l5 there may be temperature been stored for differing intervals a t differing temperatures. dependent instability at the free PSA specific epitope on the Few studies have systematically assessed the stability of PSA molecule, making free PSA less detectable at higher free and total PSA under varying storage conditions. Petters- temperatures for longer periods. The differential stability of free and total PSA under varyson et a1 examined the stability of purified PSA and PSAantichymotrypsin (ACT) complexes in vitro.14 They showed ing storage conditions has important implications for evaluthat there was no change in the concentrations of total PSA ation of the percentage of free PSA within a cancer screening and free PSA immunoreactivities when purified free PSA setting. The greater median decrease in free PSA in all storwas stored for 4 weeks at 4C or 35C. However, the concen- age conditions will necessarily decrease the percentage of tration of purified PSA-ACT complexes was not stable under free PSA. Depending on the storage conditions, the screening the same conditions. Their data suggested dissociation of the cutoffs used for total and percentage of free PSA, and the purified PSA-ACT complexes to free PSA and free ACT dur- baseline level of total and free PSA for an individual subject, ing storage with dissociation occurring more rapidly at the the decrease in percentage of free PSA as a n artifact of warmer temperature. In contrast, Chen et al reported that storage conditions may result in a n increase in false-positive free PSA-ACT complexes in the lyophilized form were stable screening tests (and unnecessary biopsies). However, it at temperatures up to 54C.18 They hypothesized that the should be noted that with the exception of storage a t tempercomplex is held together by numerous noncovalent bonds atures of 4C, the median decrease in percentage of free PSA that are unstable in aqueous solution. Thus, lyophilization or levels is within our previously reported estimate for biologifreeze-drying may be an important tool for stabilizing the cal variability of percentage of free PSA (that is 15%).11 Additionally, there are currently only estimates of optimal complex. Stenman et al measured PSA and PSA-ACT in frozen cutoffs for use of percentage of free PSA in prostate cancer serum samples collected in Finland from 1968 to 197Z7 The screening.8 Our results suggest that if frozen sera were used samples were frozen at -20C and thawed once in 1985 before in determining these cutoffs, they may be too low to apply to their study. The mean PSA immunoreactivity of 118 cases of fresh sera. Our study is unique in terms of its large sample size, frozen sera was compared to that in fresh samples from 192 randomly selected men 55 to 65 years old. Since there was inclusion of well characterized clinical groups and large a 38% lower mean in the frozen sample, they estimated quantity of sera which will allow us to examine free and total that 38% of the PSA immunoreactivity had been lost during PSA at future time points. Additionally, our study more closely mimics real life situations in which sera may be storage. Woodrum et a1 reported that free PSA was less stable than stored a t less than optimal conditions for some time prior to total PSA.1" Storing 29 fresh serum samples a t -2OC or testing. Presently, some reference laboratories provide free - 70C for approximately 1 year, these investigators found PSA measurements to patients on blood samples sent long that total PSA was stable a t these storage temperatures and distances. Therefore, reference laboratories may receive samthat 0.9% of the free PSA was lost per month of storage at ples that have not been appropriately refrigerated, and measurement of free and total PSA may not occur on the same day -2OC and 0.4% was lost per month at -7OC. Similarly, our results confirm that total PSA is relatively as the sample is drawn. Our data suggest that significant stable when stored for up to 9 months a t temperatures of amount of free PSA immunoreactivity is lost even when -20C or lower (median percent change ranged from 1 to serum is stored for as little as 24 hours at temperatures -5%, fig. 2). More specifically, the decrease in total PSA with warmer than -2OC. longer storage intervals did not significantly differ, depending on whether the serum was stored in -2OC or -7OC. For CONCLUSIONS storage intervals of 24 hours a minimal change was found for For storage intervals of up to 9 months total PSA is more total PSA regardless of storage temperature (median percent stable than free PSA under temperature conditions ranging change ranged from 0 to 1%,fig. 2). In contrast, free PSA was from 4C to -7OC. This differential stability has important less stable in general than total PSA. The percent change in implications for the clinical use of the percentage of free PSA free PSA ranged from -5 to -9% at 24 hours across all in prostate cancer screening. temperature conditions (fig. 3). For storage intervals of 2 weeks the statistical analyses indicated that storage temperREFERENCES atures of -7OC preserved stability to a greater extent than storage temperatures warmer than -70C but the actual 1. Wingo, P. A., Tong, T. and Bolden, S.: Cancer statistics, 1995. CA, 4 5 8, 1995. magnitude of the difference in percent change was relatively 2 . Jacobsen, S. J., Katusic, S. K., Bergstralh, E. J., Oesterling, J. E., modest (that is - 8%for - 2OC versus - 5% change for - 70C). Ohrt, D., Klee, G. G . , Chute, C. G. and Lieber, M. M.: Incidence Additionally, the greater variability in percent change at of prostate cancer diagnosis in the eras before and after serum -20C after 2 weeks of storage (range 1 to -50%) compared prostate-specific antigen testing. J.A.M.A., 274: 1445, 1995. with -7OC (range 1 to -20%) indicates that this difference 3. Catalona, W. J., Smith, D. S., Ratliff, T. L. and Basler, J . W.: requires further verification. There was no difference in perDetection of organ-confined prostate cancer is increased cent change in free PSA for storage temperatures of -2OC through prostate-specific antigen-based screening. J.A.M.A.. versus -7OC at 9 months. The pattern of percent change for 2 7 0 948, 1993. total and free PSA was similar across study groups. 4. Smith, D. S . , Catalona, W. J. and Herschman, J . D.: LongitudiThe mechanism for the differential stability for free versus nal screening for prostate cancer with prostate-specific antigen. J.A.M.A., 2 7 6 1309, 1996. total PSA is not known. Based on the results from Pettersson 5. Stenman, U. H., Leinonen, J., Henrik. A,. Ranniko. S., et a1,I.l we would expect that a t increasing temperature and Tuhkanen, K. and Alfthan. 0.A,: A complex between prostatestorage interval, the free PSA-ACT complex would be more specific antigen and alpha 1-antichymotrypsin is the major likely to dissociate and, therefore, result in a greater concenform of prostate-specific antigen in serum of patients with tration of free PSA. However. this pattern was not observed prostatic cancer: assay of the complex improves clinical sensiin our data. Rather, with increasing temperature and storage tivity for cancer. Cancer Res.. 51: 222, 1991. time, measurable free PSA was lost a t a greater rate than 6. Lilja. H., Christensson, A,, Dahlen, U., hlatikainen, hl. T.. total PSA. It could be that due to the surplus of ACT in Nilsson. 0.. Pcttersson. K. and Lovrren. T.: Prostate-specific serum. a oortion of the free PSA becomes comdrxed with ~-~~ antigen in serum occurs predominantly in complex with alphaC.

~~

~

~~~

I

STABILITY OF SERUM TOTAL AND FREE PROSTATE SPECIFIC ANTIGEN 1-antichyniotrypsin. Clin. Chem., 37: 1618, 1991. 7. Stennian. U. H., Hakama, M., Knekt, P., Aromaa, A,, Teppo, L. and Leinonen, J.: Serum concentrations of prostate specific antigen and its complex with alpha 1-antichymotrypsin before diagnosis of prostate cancer. Lancet, 344: 1594, 1994. 8. Catalona. W. J., Smith, D. S., Wolfert, R. L., Wang, T. J.. Rittenhouse, H. G., Ratliff, T. L. and Nadler, R. B.: Evaluation of percentage of free serum prostate-specific antigen improves specificity of prostate cancer screening. J.A.M.A., 2 7 4 1214, 1995. 9. Luderer, A. A., Chen, Y.-T., Soriano, T. F., Kramp, W. J., Carlson, G., Cuny, C., Sharp, T., Smith, W., Petteway, J., Brawer, M. K. and Thiel, R.: Measurement ofthe proportion of free to total prostate-specific antigen improves diagnostic performance of prostate-specific antigen in the diagnostic gray zone of total prostate-specific antigen. Urology, 46: 187, 1995. 10. Yuan, J. J. J., Coplen, D. E., Petros, J . A,, Figenshau, R. S., Ratliff, T. L., Smith, D. S. and Catalona, W. J.: Effects of rectal examination, prostatic massage, ultrasonography and needle biopsy on serum prostate specific antigen levels. J . Urol., 147: 810, 1992. 11. Ornstein, D. K., Rao, G. S., Smith, D. S., Ratliff, T. L.. Basler, J. W. and Catalona, W. J.: Effect of digital rectal examination and needle biopsy on serum total and percentage of free prostate specific antigen levels. J. Urol., 157: 195, 1997. 12. Tchetgen. M . B., Song, J. T., Strawderman, M., Jacobsen, S . J. and Oesterling, J. E.: Ejaculation increases the serum

2 187

prostate-specific antigen concentration. Urology. 47: 511, 1996. 13. Riehmann, M., Rhodes, P. R., Cook, T. D., Grose. G. S. and Bruskewitz, R. C.: Analysis of variation in prostate-specific antigen values. Urology, 4 2 390, 1993. 14. Pettersson, K., Piironen, T., Seppala, M., Liukkonen, L.. Christensson, A., Matikainen, M. T.. Suonpa, M.. Lovgren, T. and Lilja, H.: Free and complexed prostate-specific antigen (PSA): in vitro stability, epitope map, and development of immunofluorometric assays for specific and sensitive detection of free PSA and PSA-alpha,-antichymotrypsincomplex. Clin. Chem., 41: 1480, 1995. 15. Woodrum. D., French, C. and Shamel, L. B.: Stability of free prostate-specific antigen in serum samples under a variety of sample collection and sample storage conditions. Urology. 48: 33, 1996. 16. Smith, D. S., Bullock, A. D., Catalona, W. J. and Herschman, J. D.: Racial differences in a prostate cancer screening study. J. Urol., 156. 1366, 1996. 17. Cohen, J. and Cohen, P.: Applied Multiple RegressiodCorrelation Analysis for the Behavioral Sciences, 2nd ed. Hillsdale. New Jersey: Lawrence Erlbaum Associates, Publishers, pp. 296-297, 1983. 18. Chen, Z., Prestigiacomo, A. and Stamey. T. A,: Lyophilized PSAACT complex is stable (letter to the Editor). Clin. Chem., 4 2 1297, 1996.