[CANCER RESEARCH 45. 886-891, February 1985]

Prognostic Importance of Prostate-specific Antigen for Monitoring Patientswith Stages B2 to D1 Prostate Cancer1

Carl S. Killian,2 Norman Yang, Lawrence J. Emrich, Parida P. Vargas, Manabu Kuriyama, Ming C. Wang,Nelson H. Slack, Lawrence D. Papsidero, Gerald P. Murphy,3 T. Ming Chu, and the Investigators of theNational Prostatic Cancer Project3

Department ot Diagnostic Immunology Research and Biochemistry [C. S. K., F. P. V., M. C. W., L. D. P., T. M. C.] and Department of Biomathematics [L J. E., N. H. S.],Roswell Park Memorial Institute, Buffalo, New York 14263; Department of Urology, Gif u University, Gifu, Japan [M. K.]; and The Yang Laboratories, Bellevue, Washington[N. Y.I 98007

ABSTRACT

To evaluate the prognostic value of prostate-specific antigen

(PA) for detection of tumor growth after definitive therapy, 602sera from 70 patients with stages B2 to D! prostate cancer (26of whom recurred) were analyzed in a blind study. Using Cox's

proportional-hazards model, a highly significant association wasfound between serially measured PA and disease-free survivaltime (p = 0.0002). A positive predictive value of 100% was foundfor some markedly elevated PA levels and confirmed recurrenceof disease. In fact, this study suggested that once a PA level of88 ng/ml was reached, there was an average time of less than2 months before a recurrence was clinically confirmed. Tumorgrowth in patients who recurred was indicated by a PA elevationbefore recurrence in 92% (24 of 26) as opposed to 20% (9 of44) in disease-free patients. Additionally, in these 24 of 26patients, levels of PA were elevated 12 months (mean lead time)before a confirmed disease recurrence. In patients who were stilldisease free, serial PA appeared to increase concurrently withputative tumor growth as shown by the initial surgical stage.Generally, the greater the PA level the more advanced was thestage of disease (B2 to D,). These data suggest that PA may bea useful adjuvant marker for monitoring tumor growth in patientswith regionally confined prostate cancer.

INTRODUCTION

Prostatic carcinoma is the second most common cancer andthe third most prevalent cause of cancer deaths of men in theUnited States (27). Survival and therapeutic response are betterin patients with early stage disease than in those with metastaticcancers (24). Unfortunately, advanced disease with disseminatedmetastasis is presented in the majority of patients upon diagnosis

1This work was supported in part by Research Grants CA-15126, CA-15437,CA-29544, CA-16056, and CA-34903 awarded by the National Cancer Institute.

2To whom requests for reprints should be addressed.3G. P. Murphy is Director of the Roswell Park Memorial Institute and an

Investigator of the National Prostatic Cancer Project. Other Investigators of theNational Prostatic Cancer Project are Stefan A. Loening, University of Iowa Hospitals and Clinics; Robert P. Gibbons, The Mason Clinic; George R. Prout, Massachusetts General Hospital; William W. Scott, Johns Hopkins Hospital; Mark S.Soloway, University of Tennessee; Joseph D. Schmidt, University of California atSan Diego; Raju Thomas, Tulane Medical Center; James M. Pierce, Hutzel Hospital/Wayne State University; Peter T. Scardino, Baylor University; David G. McLeod,Walter Reed Army Medical Center; Charles F. McKiel, Rush-Presbyterian/St. Luke's

Medical Center; Jean B. deKemion, University of California at Los Angeles; J.Edson Pontes, Roswell Park Memorial Institute; John H. Lynch, GeorgetownUniversity; Robert C. Flanigan, University of Kentucky; Jackson E. Fowler, University of Virginia School of Medicine; Hugh Fisher, Albany Medical Center; Robert M.Weissman, University of North Carolina at Chapel Hill; Joseph C. Cemy, HenryFord Hospital; John Trachtenberg, Toronto General Hospital; and Fernand Labrie,Laval University Medical Center.

Received May 22,1984; accepted November 1, 1984.

(9). Consequently, reliable biological markers for diagnosis andprognosis in these patients have been sought for over 5 decades(1-3, 5, 7, 10, 13, 14, 25, 31, 33). A new candidate for follow-

up and assessment of therapeutic efficacy in prostate cancerpatients is PA4 recently identified by Wang ef al. (28-30) of our

group.A primary reason that suggests PA as a useful marker of

prostate cancer is that PA is cell-type and organ-site specific.Previous studies within our group as well as from other investigators have shown that PA is localized only within epithelia ofprostatic ductal cells (19, 21, 28). PA has been demonstrated byNadji et al. (19) and Papsidero et al. (20) to be a useful histologicalprobe for the detection and differential diagnosis of metastaticcancer when combined with specific antibody directed againstPA. Several studies by our group in conjunction with the NationalProstatic Cancer Project (NPCP) have also shown PA to beuseful for monitoring treatment response for prostatic cancerpatients (17). Kuriyama ef al. (16) using a sensitive enzyme-linkedimmunosorbent assay have produced data from advanced stage(D2) patients that strongly suggest a prognostic relationshipbetween pretreatment levels of serum PA and patients' survival.

Pontes ef al. (23) using serial PA levels from Stages B2 to Dipatients also found an excellent correlation between an elevatedserum PA and the development of metastatic disease in theprostatectomized group.

This study was initiated to evaluate, in a blind manner, theprognostic effectiveness of PA for detection of tumor growthafter definitive therapy in patients with regionally confined prostate cancer. The end point of tumor growth was a clinicallydetectable recurrence of disease in clinically disease-free patients. The prognostic importance of PA for monitoring thesepatients was statistically analyzed using Cox's (6) proportional

hazards model. Using serially measured PA levels, the significance of low PA levels relative to high PA levels in terms of alonger disease-free survival time in patients was evaluated. This

analysis also adjusted for the effects of stage and therapy ondisease-free survival time. Additionally, the usefulness of PA for

monitoring these patients is illustrated using some commonparameters for tumor marker evaluation, such as predictivevalue, time to recurrence, mean lead time before recurrence, andlongitudinal patient evaluation.

MATERIALS AND METHODS

Specimens. Six hundred two serial specimens were collected over a6-year period from 70 patients with regionally confined prostate cancer.

4The abbreviations used are: PA, human prostate-specific antigen; NPCP,

National Prostatic Cancer Project; UNL, upper normal limit; xUNL, upper normallimit of assay x PA concentration; RIA, radioimmunoassay; PAP, prostatic acidphosphatase.

CANCER RESEARCH VOL. 45 FEBRUARY 1985

886

Research. on September 13, 2020. © 1985 American Association for Cancercancerres.aacrjournals.org Downloaded from

PROGNOSTIC VALUE OF PA

Patients were surgically staged on the basis of lymph node dissectionand other evaluations (11, 32). Eleven of the patients were Stage B2,14were Stage C, and 45 were Stage DL Patients having initially receivedeither radical prostatectomy, cryosurgery, or radiation therapy werereceiving long-term adjuvant chemotherapy of Cytoxan or Estracyt or noadditional treatment during the evaluation period. Follow-up was performed at 6-week intervals. Twenty-six of the 70 patients presentedclinical recurrence of disease during follow-up as confirmed by radi

ographie and radionuclide examinations as well as by other clinical testsand subsequently received additional therapy in advanced disease protocols (26). Basis for patient selection was sera availability. Sera frompatients with a minimum of 5 serial specimens available were used forPA analysis by one of us (N. Yang). No prior knowledge of the clinicalstatus of patients or sera was communicated. PA analyses by a double-

antibody RIA were completed within 3 days after receipt of specimens.PA Analysis. Purification of PA, production of anti-PA sera, and

specificity of the antisera have been reported previously (30). Thesereagents were used for PA analysis. The quantitation of serum PA wasperformed using a standard double-antibody RIA method (18). Briefly,labeled PA, PA standards, and PA in serum competed for rabbit anti-PA

in test tubes. Tubes were incubated for 24 hr at room temperature. Goatanti-rabbit serum was added, and after a 1-hr incubation at room tem

perature, the test tubes were centrifugea and radioactivities in theprecipitate were counted. A standard PA curve was linear between 0.5to 50.0 ng/ml using the logit-log plot of percentage bound divided by

percentage bound at 0 concentration versus PA concentration. Coefficients of variation for 20 analyses each of 2, 5, and 8 ng/ml in interassayquality control were 9.0,6.3, and 5.9%, respectively; for ¡ntrassayqualitycontrol, the same 3 levels of PA were 8.6, 5.3, and 4.2%, respectively.A mean and S.D. of 0.75 and 0.71 ng/ml, respectively, were obtainedfrom 127 apparently healthy males. In 38 sera from normal females, all"PA" levels were less than 0.5 ng/ml, i.e., the sensitivity level of the

assay. Using a mean + 2 S.D., a normal range for PA by the RIA methoddescribed, was 0 to 2.2 ng/ml. This value was similar to the reportednormal range of our enzyme immunoassay, 0 to 1.8 ng/ml (15, 17).Serum PA level (mean + 2 S.D.) from 12 patients with benign prostatichyperplasia was 12.74 ng/ml, higher than 7.5 ng/ml as obtained previously by enzyme immunoassay. Since patients with benign prostatichyperplasia were not the subject of this study, UNL for PA was acceptedas 2.2 ng/ml, as all patients had been diagnosed with prostate cancer,had received definitive therapy, and were being evaluated during thefollow-up intervals.

Methods of Prognostic Evaluation. The prognostic value of seriallymeasured levels of PA for predicting a clinically detectable recurrence ofdisease during long-term follow-up in patients with regionally confined

prostate cancer was evaluated by the following methods.Cox's proportional-hazards model with time-dependent covariates

was used to analyze the relationship between the serially measured PAlevels and disease-free survival time, defined as the time from definitive

therapy to clinically detectable recurrence of disease. For this method,the PA level was assumed to remain unchanged between serially évaluable measurements. Forty-three patients, of whom 15 have recurred,

were used for these analyses. These were patients with at least one PAmeasurement within 12 weeks after definitive therapy and who receivedsufficient therapy and presented no protocol violations.

Positive predictive value (8) of a single elevated PA level for diseaserecurrence was estimated, irrespective of therapy and time of follow-up,

using the most elevated PA level in serial measurements from each of26 patients with confirmed disease recurrence and 44 disease-free

patients. The percentage of patients who recurred with PA values at aspecific level of PA to the total number of patients with these values atthe PA level was used to determine the positive predictive value. Inaddition, the time to recurrence was compared with the magnitude of asingle elevated PA value. The highest PA value in serial specimens fromindividual patients that recurred was used to determine the time fromthat PA result to recurrence. A mean lead time to recurrence versus

magnitude of PA was calculated.Serial levels of PA were plotted for patients with a clinically detectable

disease recurrence and for patients who were clinically disease free. Thescattergram included all PA levels within a 1-year period prior to recurrence in patients who had recurred, and all PA levels within a 1-year

period prior to the last measured PA level in patients who were stilldisease free.

To determine whether that PA may elevate prior to a confirmed diseaserecurrence, the time between the first of a series of PA values greaterthan the UNL or a single elevated value greater than 5 times the UNLwas determined for the patients who had recurred.

Serial levels of PA were longitudinally evaluated with disease statusfor monitoring effectiveness in patients with and without confirmeddisease recurrence. PA was studied in terms of (a) elevation beforedisease recurrence, (b) normalization after recurrence when additionaltherapy was received by a patient who was placed in advanced diseaseprotocols of NPCP, (c) normalization of PA levels in patients withoutdisease recurrence, and (d) relationship of serial PA levels with initialstage of disease.

RESULTS

PA and Disease-free Survival Time. Statistical evaluationusing Cox's proportional hazards model showed a highly signif

icant relationship between serially measured PA and disease-free survival time (p = 0.0002, Table 1). The coefficient estimates

and p listed in Table 1 reveal the following: if a coefficient issignificantly different from zero, then the covariate correspondingto that coefficient is an important prognostic variable. The significantly positive coefficient for PA levels indicates a longer disease-free survival time for patients with low PA levels than for

patients with high levels. The coefficients corresponding to thecovariates "Stage C" and "Stage D/' indicate the risk of recur

rence for patients with Stages C and Di, respectively, relative topatients with Stage B2. The covariate "Radiotherapy" is an

indicator variable for initial treatment; the coefficient for thiscovariate indicates the risk of recurrence for patients who received radiotherapy, relative to patients who received prostatectomy or cryosurgery. Similarly, the coefficient for the covariate"Estracyt" indicates the risk of recurrence for patients receiving

Estracyt adjuvant therapy relative to patients not receiving Estracyt. These covariates were included in the model to adjust forthe effect of stage and treatment on disease-free survival time.

As indicated in Table 1, no significant differences were found forcovariates other than PA.

Positive Predictive Value of a Single Elevated PA Result. Asingle elevated level of PA was evaluated for prediction ofdisease recurrence. The positive predictive values shown inChart 1 were determined using the most elevated PA level inserial specimens of each patient with and without a confirmed

Table 1Association between disease-free survival time and PA

Covariates inmodel*Stage

CStage D,RadiotherapyEstracytPA"Coefficient

estimates-1.107

0.2280.763

-0.550

2.392P0.22

0.420.130.250.0002

' Cox's proportional hazards model." Marker measurements were transformed to tog«(10+ measurement) for the

analysis.

CANCER RESEARCH VOL. 45 FEBRUARY 1985

887

Research. on September 13, 2020. © 1985 American Association for Cancercancerres.aacrjournals.org Downloaded from

PROGNOSTIC VALUE OF PA

6/6

40

PA Level Times Upper Normal LimitChart 1. Positive predictive value of PA for disease recurrence. Positivepredic

tive valueequals the percentageof patients who recurred with most elevatedserialPA value greater than indicated PA level to total number of patients with mostelevated serial PA value greater than indicated PA level. Positive predictive valueincreaseddirectly with increasedlevel of PA.

31 40

PA Level Times Upper Normal LimitChart 2. PAversus time to recurrence.The highest PAvalue in serialspecimens

from individualpatients that recurred was used to determine the time from that PAmeasurement to a confirmed recurrence. Shown are number of months to recurrence [mean ±S.D. (oars)) for patients who recurred at a level of PA between 0and 1, >1 and 5, >5 and 20, >20 and 31, >31 and 40, and above. Months toclinicallydetectable recurrence decreased inverselywith increased level of PA.

disease recurrence. The percentage of patients who recurredand presented PA values at or above the specific level shown tothe total number of patients that presented PA values at theselevels was evaluated as the positive predictive value of PA fordisease recurrence at the specific level of elevation. As shownin Chart 1, positive predictive values of PA for disease recurrenceincreased concurrently with increased levels of PA. A positivepredictive value of 100% was obtained for disease recurrence inpatients presenting an elevated PA level of 40 xUNL. Interestingly, even an elevation of PA at the 5 xUNL level (11.0 ng/ml)correctly predicted a disease recurrence in 72% of patients.

In addition, time to recurrence was compared with magnitudeof a single elevation of PA. An inverse relationship was shownbetween an amount of time before a disease recurrence wasclinically detectable and a level of PA. As shown in Chart 2, anaverage time of less than 2 months was obtained for a PA levelof 40 x UNL with 100% (6 of 6) of patients at this PA level havingpresented a confirmed disease recurrence, whereas in 6 patientswith a PA level within a range of 1 to 5 xUNL, an average timeof 7 months was found before a disease recurrence was clinicallydetectable. Time to recurrence (mean ±S.D.) at a specific levelof PA was estimated using the highest PA values in serialspecimens from individual patients and determining the time

required from that PA result to a confirmed disease recurrenceof the patient.

Elevation of PA prior to Disease Recurrence. PA levels inpatients with disease recurrence and who were disease free areillustrated in Chart 3. The data points shown in the figure denoteall serial levels of PA within 12 months prior to a clinicallydetectable disease in patients who had recurred and all seriallevels 12 months before and including the last PA measurementin patients that were clinically disease free. Twenty-four of 26

patients that developed disease recurrence presented PA valuesgreater than the upper normal limit, as opposed to 9 of 44 amongdisease-free patients. Also, in 15 of 26 patients with detectable

disease recurrence, all serial levels of PA were presented abovethe upper normal limit as compared with 5 of 44 patients in thedisease-free group.

Mean Lead Time of PA Elevation prior to Recurrence. Themean lead time of elevation in PA level before a detectabledisease recurrence was ascertained for patients with regionallyconfined prostate cancer. The time between the first of a seriesof elevated PA values, or a single elevated PA level (>5 xUNL),and a confirmed disease recurrence was used to determine themean lead time. The mean lead time for 26 patients ranged from1 to 23 months (mean = 12 ±11 (S.D.) months) overall. A single

elevated PA level of >5 xUNL was used as a significant numberin recurring patients with only one PA measurement beforedisease recurrence, since 72% of all patients with a PA value atthis level had recurred.

Longitudinal PA and Disease Status of Prostate CancerPatients. Serial levels of PA were longitudinallyevaluated withdisease status for monitoring effectiveness in patients with and

<

z

COu

wo

(TU

oooooooooo

000000000000

oooooooooooooooooooo

oooooooooooooooooooooooooooooo

CLINICALLY DISEASE FREE CLINICAL DETECTABLE RECURRENCE

Chart 3. Serial levels of PA in patients with and without a clinical diseaserecurrenceafter receivinga curative therapy for regionallyconfinedprostate cancer.Data points illustrate all serial levels of PA within 12 months prior to a diseaserecurrence (recurring group) or all serial values within 12 months before andincluding the last serial results (disease-freegroup). In the recurrence group, 24 of26 patients presented serial levels greater than the upper normal limit ( ) asopposed to 9 of 44 disease-freepatients.

CANCER RESEARCH VOL. 45 FEBRUARY 1985

888

Research. on September 13, 2020. © 1985 American Association for Cancercancerres.aacrjournals.org Downloaded from

PROGNOSTIC VALUE OF PA

without confirmed disease recurrence. Representative patternsof serial PA levels in patients with clinically detectable recurrenceof disease are shown in Chart 4. In general, and also as indicatedin Table 1 and Charts 2 and 3, the greater the level of PA thesooner a confirmed disease recurrence was presented in patientswho recurred. Serial levels of PA were generally increasinglyelevated or remained elevated from initiation of study to time ofa clinically detectable disease recurrence. After disease recurrence, serial PA levels generally normalized or were slightlyelevated when patients were receiving additional therapy inadvanced disease protocols of the NPCP.

Chart 4 illustrates 4 typical patterns shown by 24 of 26 patientswho recurred and presented serial levels of PA that were elevated above the normal limit, before or during recurrence. Asdepicted, some patients presented patterns showing serial levelsof PA that initially were low (Chart 4, Pattern A) or were near theupper normal range (Chart 4, Patterns B and C), and then eitherbecame increasingly elevated before a recurrence (Chart 4,Patterns A and C) or decreased to within the normal range for aperiod of time and then became increasingly elevated before arecurrence (Chart 4, Pattern B). Some patients that initiallyshowed highly elevated PA levels after definitive therapy werefound to present a confirmed disease recurrence a few monthslater (Chart 4, Pattern D). In 13 of 15 patients whose PA valueswere available after a disease recurrence, serial levels of PAdecreased to within or slightly above the normal range aftertherapies for advanced disease were given (Chart 4, Patterns B,

too

60

Chart 4. Longitudinal follow-up of 4 patterns illustrating the trend of PA ob

served in 26 patients with a confirmed disease recurrence (fl) (a) after definitivetherapy. Serial levels of PA were increasingly elevated before (• •)or duringrecurrent disease detection in 24 of 26 patients. After definitive therapy, severalpatients presented Pattern A where serial levels of PA were initially low and graduallyincreased to above the upper normal range ( ) before a disease recurrence.More typically observed patterns shown by patients were serial PA levels presentedinitially near the upper normal range and then either became increasingly elevatedbefore recurrence (C) or decreased to within the normal range for a period of timeand then became increasingly elevated before recurrence (B). Some patients whopresented initial highly elevated levels of PA after definitive therapy were found tohave recurred a few months later (D). In 15 patients, all serial levels of PA werepresented above the normal range. In 13 of 15 patients whose PA values wereavailable after recurrence (• •),serial levels of PA decreased to within orslightly above the normal range when patients were receiving additional therapy inadvanced disease protocols (Patterns B, C, and D). In general, the greater the levelof PA, the shorter the time before clinical detection of a disease recurrence. Bj, C,D,, stage of disease.

C, and D). In contrast, 27 of 44 disease-free patients presented

all levels of PA that were within the normal range. In 12 of 44patients, PA levels were slightly elevated after initial definitivetherapy but subsequently were within the normal range or slightlyelevated during follow-up. The remaining 5 disease-free patients

presented all serial levels of PA elevated above the upper normallimit, at the time of this analysis.

Although not statistically significant, Table 2 shows an association between serial levels of PA and the initial surgical stageof disease in clinically disease-free patients. Ranges of serial PA

as well as means and S.D.s appeared to increase concurrentlyas "occult" tumor growth increased. All Stage B2 patients were

within normal limits, as were most Stage C patients, whileapproximately 40% of the Stage D, patients were at least initiallyabove the upper normal limit.

DISCUSSION

PA is a recently identified biological marker for prostate cancer(30). PA is a chemically and ¡mmunologically well-defined glyco-

protein and distinct from PAP (28,29). Interestingly, PA has beenfound to be more significant than is PAP for monitoring patientswith prostate cancer. Our preliminary statistical results (data notshown) have revealed that PA and PAP both are of value;however, PA is shown to be prognostically more useful for patientfollow-up. Although PA has not been found to be efficient for

general screening of prostate cancer, PA has been establishedas a marker for human prostatic epithelial cells (19, 21, 22), asan immunohistological marker for prostate neoplasms (20, 22),and as a sérodiagnostic prognosticator for patients with advanced prostatic cancer (15, 23).

A reliable marker, among other requirements, should predictan increased growth of malignant cells before a tumor hasprogressed to clinical symptomology. Immunohistochemicalstudies have shown that PA is expressed on both primary andmetastatic prostate tumors (19, 21). Thus, serum PA levels mayincrease concurrently with an increased tumor volume. A suitablemodel to study the reliability of PA as a marker of tumor growthis provided in this study by clinically disease-free patients after adefinitive "curative" therapy for regionally confined prostate can

cer. This model should offer useful information about PA andtumor growth, because these patients, in theory at least, arewithout viable prostatic cancer and therefore should be trulydisease free.

Since a cancer is seldom detectable clinically before a minimumtumor volume of 1 cm diameter has occurred (4), a confirmeddisease recurrence in disease-free patients would indicate tumor

growth. Therefore, in this study, the prognostic value of serial

Table2Serial levels of serum PA and initial surgical stage

Stage'B,C

D,No.

ofpatients/no.measurements9/72

13/11722/202PA

level(XUNL)Range0.02-0.50

0.11-4.090.10-38.18Mean

±S.D."0.22

±0.170.69 ±1.052.43 ±4.02PcNS"

NSNS

" Surgical stages in clinicallydisease-freepatients during follow-up.6 Group mean ±S.D. of individual mean for serial values in patients.c Student's f test after transformation of PA measurementto log,0.d NS, not significant.

CANCER RESEARCH VOL. 45 FEBRUARY 1985

889

Research. on September 13, 2020. © 1985 American Association for Cancercancerres.aacrjournals.org Downloaded from

PROGNOSTIC VALUE OF PA

PA levels were statistically evaluated for disease-free survival

times. In addition, the usefulness of PA for detection of tumorgrowth before the clinical manifestation of a disease recurrencewas illustrated with commonly used immunodiagnostic parameters.

The significant prognostic value of serially measured PA fordisease-free survival time, as evaluated using Cox's proportional-

hazards model, suggests that PA is an important marker inpatients' follow-up (Table 1). One should keep in mind, when

reviewing the results of this analysis, that the PA level wasassumed to remain unchanged between serially évaluablemeasurements. A precise analysis would require a daily measurementof PA for each patient. Our analysis, however, did incorporatean adjustment for the effects of disease stage and treatmentreceived as indicated while evaluating PA with disease-free sur

vival time. Interestingly, PA was found to be a prognosticallyimportant marker (Table 1) even after adjustment for the effectof Estracyt on disease-free survival time. The lack of apparent

estrogenic effects following Estracyt administration to humanshas been reported previously (12).

Should a serum PA level increase concurrently with an increasein tumor load, it is reasonable to expect that at some elevatedlevel PA would predict a recurrence of disease. Galen andGambino (8) showed that the predictive value model is an effective method to evaluate the usefulness of laboratory diagnostictests. Therefore, we have evaluated the positive predictive valueof a single elevated level of PA as one of the criteria for diseaserecurrence. Although the positive predictive value decreased withdecreasing level of PA, results shown in Charts 1 and 2 indicate,e.g., that a PA level of 40 xUNL (88 ng/ml) would require anaverage time of less than 2 months before a recurrence wasclinically confirmed. The use of time to recurrence for a specificlevel of PA was not intended in this study as an absolute periodof time but was used to illustrate an association between PAlevel and tumor growth. Our results showed that the greater anelevation of PA, the higher a positive predictive value for PA wasfound and the earlier a progression of disease was confirmed.However, as the marker is used to predispose a disease progression, an elevated PA must be evaluated against the positivepredictive value of marker similar to other clinical laboratoryresults. Of course, one should not suggest the use of a singlymeasured value of any biological marker for patient care withoutverification of marker level and other laboratory and clinicalevidence for corroboration.

It is important that a tumor marker indicates tumor growthbefore clinical detection occurs. Our results showed that PAlevels were elevated between 0 and 23 months (mean, 12months) before a confirmed disease recurrence. When all serialvalues in both patient groups were compared for PA elevationwithin a 1-year period prior to either confirmed recurrence or lastPA results (disease-free group), the PA levels were shown to be

markedly elevated in patients who had recurred, as comparedwith the levels in patients who had not recurred (Chart 3).Although 2 patients in the recurrence group failed to present anelevated level of PA before or at the time of recurrence, 92% ofpatients with recurrence did present at least one elevated PAlevel as opposed to 20% of the patients who were "still" disease

free. Also, in 58% of the former group, all serial levels of PAwere elevated as compared with 11% of patients in the lattercategory. These data, in general, suggest that serial PA levels

increase with tumor growth, as disease-free patients later de

velop a clinical recurrence of disease.The prognostic value of PA may be appreciated more fully

using longitudinal evaluation of PA levels and disease status ofboth patient groups. In patients with recurrence (Chart 4), datashowed that serial levels of PA were elevated before a confirmeddisease recurrence, normalized or decreased to slightly elevatedlevels after a recurrence in a majority of patients who werereceiving additional therapy in advanced disease protocols, andthus were suggestive of favorable response to new therapy. Inthe disease-free group, a majority of patients showed serial levelsthat remained within the normal limit for long periods of disease-free response, while some patients presented PA levels thatwere initially elevated, then subsequently decreased to withinnormal limits. It was not surprising that some (5 of 44) disease-free patients have presented all serial levels of PA elevatedabove the UNL, since PA has been shown to elevate before aconfirmed disease recurrence. One of these patients, however,presented all serial PA results between 10 to 30 xUNL withouta confirmed disease recurrence. No explanation can be offeredat this time for these highly elevated levels without a diseaserecurrence, unless residual viable tumor was present. In fact,serial levels of both PAP and acid phosphatase also were allelevated 5 to 6 and 2 to 3 times their UNL, respectively. Unfortunately, this patient has withdrawn from study, and no furtherevaluation is possible. However, other disease-free patients will

continue to be followed.In conclusion, the reliability of PA as a marker was evaluated

in a blind study for detection of tumor growth before a confirmeddisease recurrence after definitive therapy in patients with regionally confined prostate cancer. Highly elevated PA levels arestrongly suggestive of disease recurrence. Levels of PA mayelevate before confined disease recurrence. Levels of PA mayremain within a normal range during a disease-free period or

after a favorable response to additional therapy. These datasuggest that PA may be a useful adjuvant marker for monitoringtumor growth in patients with regionally confirmed prostatecancer, and perhaps with further statistical study, highly elevatedlevels of PA may suggest the exclusion of patients believed tohave clinically localized prostate cancer from receiving definitivetherapy.

ACKNOWLEDGMENTS

The authors thank Mark Brady, Jacqueline Hoges, and Brenda Shelley fortechnicalsupport; ElenaGrecoand GeorgeFox for illustrativeassistance;and JoanM. Ogledzinski for manuscript preparation.

REFERENCES

1. Ablin, R. J., Branson, P., Soanes,W. A., and Witebsky, E. Tissue and species-specific antigens of normal human prostatic tissue. J. Immunol., 704: 1329-1339, 1970.

2. Chu, T. M., Wang, M. C., Lee, C. L., Killian,C. S., Valenzuela,L. A., Wajsman,Z., Slack, N. H., and Murphy, G. P. Enzyme markers for prostatic cancer.Cancer Detect. Prev., 2: 693-706,1979.

3. Chu, T. M., Wang, M. C., Scott, W. W., Gibbons, R. P., Johnson, D. E.,Schmdit, J. D., Leoning, S. A., Prout, G. R., and Murphy, G. P. Immunodetec-tion of serum prostatic acid phosphatase. Methodologyand clinicalevaluation.Invest. Urd., 15: 319-322, 1978.

4. Collins, V. P., Loeffler, R. K., and Tivey, H. Observations on growth rates ofhuman tumors. Am. J. Roentgenol. Radium Ther. Nucí.Med., 76: 988-1000,1956.

5. Cooper,J. F., and Foti, A. A radioimmunoassayfor prostatic acid phosphatase.I. Methodology and range of normal male serum values. Invest. Ural., 12: 98-

CANCER RESEARCH VOL. 45 FEBRUARY 1985

890

Research. on September 13, 2020. © 1985 American Association for Cancercancerres.aacrjournals.org Downloaded from

PROGNOSTIC VALUE OF PA

102, 1974.6. Cox, D. R. Regression models and life tables. J. R. Statistical Soc. Ser. B.,

34:187-200, 1972.7. Flocks, R. H., Urich, V. C., Patel, C. A., and Opitz, J. M. Studies on the antigen

properties of prostate tissue. I. J. Urol., 8: 134-143, 1960.8. Galen, R. S., and Gambino, S. R. Beyond Normality. The Predictive Value and

Efficiency of Medical Diagnoses. New York, NY: John Wiley & Sons, Inc.,1975.

9. Gittes, R. F., and Chu, T. M. Detection and diagnosis of prostatic cancer.Semin. Oncol., 2: 123-130,1976.

10. Gutman, E. B., Sproul, E. E., and Gutman, A. B. Significance of increasedphosphatase activity of bone at the site of ostéoplastiemétastasessecondaryto carcinoma of the prostate gland. Am. J. Cancer, 28: 485-496,1936.

11. Harmer, M. H. (ed.). TNM classification of malignant tumors. In: UICC International Union Against Cancer, Ed. 3, pp. 118-133. Geneva: WHO, 1980.

12. Kadohama, N., Kirdani, R. Y., Murphy, G. P., and Sandberg, A. A. Estramustinephosphate: metabolic aspects related to its action in prostatic cancer. J. Urol.,119: 235-238, 1978.

13. Killian, C. S., Vargas, F. P., Pontes, E. J., Beckley, S., Slack, N. H., Murphy,G. P., and Chu, T. M. The use of serum isoenzymes of alkaline and acidphosphatases as possible quantitative markers of tumor load in prostatecancer. Prostate, 2: 187-206,1981.

14. Killian, C. S., Vargas, F. P., Slack, N. H., Murphy, G. P., and Chu, T. M.Prostate-specific acid phosphatase versus acid phosphatase in monitoringpatients with prostate cancer. Ann. N. Y. Acad. Sci., 390:122-132,1982.

15. Kuriyama, M., Wang, M. C., Lee, C. L., Killian, C. S., Papsidero, L. D., Inaji,H., Loor, R. M., Lin, M. F., Nishiura, T., Slack, N. H., Murphy, G. P., and Chu,T. M. Multiple marker evaluation in prostate cancer using tissue specificantigens. J. Nati. Cancer Inst., 68: 99-105, 1982.

16. Kuriyama, M., Wang, M. C., Lee, C. L., Papsidero, L. D., Killian, C. S., Inaji H.,Slack, N. H. Nishiura, T., Murphy, G. P., and Chu, T. M. Use of human prostate-specific antigen in monitoring prostate cancer. Cancer Res., 41: 3874-3876,1981.

17. Kuriyama, M., Wang, M. C., Papsidero, L. D., Killian, C. S., Shimano, T.,Valenzuela, L. A., Nishiura, T., Murphy, G. P., and Chu, T. M. Quantitation ofprostate-specific antigen in serum by a sensitive enzyme ¡mmunoassay.Cancer Res., 40: 4658-4662, 1980.

18. Uedtke, R. J., and Batjer, J. D. Measurement of prostate-specific antigen byradioimmunoassay. Clin. Chem., 30: 649-652,1984.

19. Nadji, M., Tabei, S. Z., Castro, A., Chu, T. M., Murphy, G. P., Wang, M. C.,

and Morales, R. Prostate-specific antigen: an immunohistologic marker forprostatic neoplasms. Cancer (Phila.), 48:1229-1232,1981.

20. Papsidero, L. D., Croghan, G. A., Wang, M. C., Kuriyama, M., Johnson, E. A.,Valenzuela, L. A., and Chu, T. M. Monoclonal antibody (F5) to human prostateantigen. Hybridoma, 2: 139-147,1983.

21. Papsidero, L. D., Kuriyama, M., Wang, M. C. Horoszewicz, J. S., Leong, S.S., Valenzuela, L. A., Murphy, G. P., and Chu, T. M. Prostate antigen: a markerfor human prostatic epithelial cells. J. Nati. Cancer. Inst., 66: 37-41, 1981.

22. Papsidero, L. D., Wang, M. C., Kuriyama, M., Horoszewicz, J. S., Leong, S.S., Croghan, G. A.. Valenzuela, L. A., Murphy, G. P., and Chu, T. M. Humanprostate epithelial cell-specific antigen. In: G. P. Murphy, A. A. Sandberg, andJ. P. Karr (eds.), The Prostatic Cell: Structure and Function, Part A, pp. 435-443. New York: Alan R. Uss, 1981.

23. Pontes, J. E., Chu, T. M., Slack, N. H., Karr, J., and Murphy, G. P. Serumprostatic antigen measurement in localized prostatic cancer: correlation withclinical course. J. Urol., 728: 1216-1220, 1982.

24. Robbins, G. L. Male genital tract. In: Pathologic Basis of Disease, pp. 1171-1200. Philadelphia: W. B. Saunders Co., 1974.

25. Romas, N. A., Hsu, K. C., Tomashefsky, P., and Tannenbaum, M. Counter-immunoelectrophoresis for detection of human prostatic acid phosphatase.Urologia, 72:79-83, 1978.

26. Schmidt, J. D., Scott, W. W., Gibbons, R. P., Johnson, D. E., Prout, G. R.,Leoning, S. A., Soloway, M., deKernion, J., Pontes, J. F., Slack, N. H., andMurphy, G. P. Chemotherapy program of the National Prostatic Cancer Project.Cancer (Phila.), 45: 1937-1946, 1980.

27. Silverberg, E. Cancer statistics, 1982. Ca—Cancer J. Clin., 32: 15-31,1982.28. Wang, M. C., Kuriyama, M., Papsidero, L. D., Loor, R. M., Valenzuela, L. A.,

Murphy, G. P., and Chu, T. M. Prostate antigen of human cancer patients.Methods Cancer Res., 79:179-197,1982.

29. Wang, M. C., Loor, R. M., U, S. L, and Chu, T. M. Physico-chemicalcharacterization of prostate antigen purified from human prostate gland andseminal plasma. IRCS Med. Sci., 77: 327-328, 1983.

30. Wang, M. C., Valenzuela, L. A., Murphy, G. P., and Chu, T. M. Purification ofa human prostate specific antigen. Invest. Urol., 77:159-163,1979.

31. Watson, R. A., and Tang, D. M. The predictive value of prostatic acid phosphatase as a screening test for prostatic cancer. N. Engl. J. Med., 303: 497-499,1981.

32. Whitmore, W. F., Jr. The rationale and results of ablative surgery for prostaticcancer. Cancer (Phila.), 76: 1119-1122,1963.

33. Yam, L. T. Clinical significance of the human acid phosphatase. A review. Am.J. Med., 56: 606-616,1974.

CANCER RESEARCH VOL. 45 FEBRUARY 1985

891

Research. on September 13, 2020. © 1985 American Association for Cancercancerres.aacrjournals.org Downloaded from

1985;45:886-891. Cancer Res   Carl S. Killian, Norman Yang, Lawrence J. Emrich, et al.  

Prostate Cancer1 to D2Monitoring Patients with Stages BPrognostic Importance of Prostate-specific Antigen for

  Updated version

  http://cancerres.aacrjournals.org/content/45/2/886

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/45/2/886To request permission to re-use all or part of this article, use this link

Research. on September 13, 2020. © 1985 American Association for Cancercancerres.aacrjournals.org Downloaded from