OctMbmr0iR1rwipCmth

Dpeaua

HSsZF

o

oSt

1

ADULT UROLOGY

9

PROSTATE-SPECIFIC ANTIGEN VELOCITY ACCURATELYPREDICTS RESPONSE TO SALVAGE RADIOTHERAPY

IN MEN WITH BIOCHEMICAL RELAPSE AFTERRADICAL PROSTATECTOMY

RUPA PATEL, HERBERT LEPOR, ROBERT P. THIEL, AND SAMIR S. TANEJA

ABSTRACTbjectives. To determine whether prostate-specific antigen (PSA) velocity (PSAV), used as a selection

riterion for salvage radiotherapy (RT) after radical prostatectomy (RP), predicts the likelihood of responseo RT in men with biochemical relapse.ethods. We retrospectively reviewed the records of 48 patients who had undergone salvage RT for

iochemical relapse after RP. All men were followed up with serial PSA measurements for a minimum of 6onths from their initial PSA recurrence, and RT was only offered to those patients with a serum PSA level

emaining at less than 1.0 ng/mL. The response to RT was defined as maintenance of a PSA level of less than.1 ng/mL. The pathologic and clinical parameters, including PSAV, were examined to determine their

ndividual ability to predict the response to RT.esults. Of the 48 patients, 30 had maintained a PSA level of less than 0.1 ng/mL at a median follow-up of6 months. The PSAV was strongly predictive of the likelihood of a response to salvage RT. The medianelapse-free survival time for patients with a PSAV of less than 0.035 ng/mL/mo was 28 months comparedith 16 months for patients with a PSAV greater than 0.035 ng/mL/mo. All other parameters tested,

ncluding Gleason score, seminal vesicle invasion, extracapsular extension, and margin status, were notredictive of the likelihood of a response to RT.onclusions. In the present study, PSAV accurately predicted the likelihood of response to salvage RT inen with biochemical relapse after RP. No other pathologic parameters predicted the likelihood of response

o RT. Using PSAV as a sole selection criterion for salvage RT after RP may allow improvement in theistorically low rates of durable response. UROLOGY 65: 942–946, 2005. © 2005 Elsevier Inc.

bcrwr

obhpwgglPt

espite the early detection of prostate cancer,20% to 35% of men undergoing radical

rostatectomy (RP) for clinically localized dis-ase will demonstrate a rising prostate-specificntigen (PSA) level within 5 years of follow-p.1– 4 A PSA rise, termed biochemical relapse,fter RP may be due to local, metastatic, or com-

. Lepor is co-owner of MedReviews LLC, New York, New York;. S. Taneja is a consultant and speaker and received researchupport from Bayer Diagnostics and is a speaker for Astra-eneca; and R. P. Thiel is a consultant for Bayer Diagnostics,ujirebio, Abbott Laboratories, and Roche Diagnostics.From the Department of Urology, New York University School

f Medicine, New York, New YorkReprint requests: Samir S. Taneja, M.D., Department of Urol-

gy, New York University School of Medicine, 150 East 32ndtreet, Suite 200, New York, NY 10016. E-mail: samir.aneja@nyumc.org

Submitted: June 24, 2004, accepted (with revisions): December

l, 2004

© 2005 ELSEVIER INC.42 ALL RIGHTS RESERVED

ined recurrence. Because only isolated local re-urrence is amenable to cure after external beamadiotherapy (RT), the evaluation of patientsith biochemical relapse should be targeted to

ule out metastatic disease.To date, several investigators have demonstrated

verall long-term rates of response to salvage RT toe approximately 10% to 62%.5–14 Several reportsave noted an increased likelihood of metastaticrogression after biochemical relapse among thoseith seminal vesicle invasion, Gleason scorereater than 7, capsular penetration, positive sur-ical margins, or recurrence within 1 year of fol-ow-up.6,9,12,14,15 Others have correlated the pre-RTSA levels with the likelihood of durable responseo salvage RT.11–13,15

Several investigators have also reported PSA ve-

ocity (PSAV) to be a strong predictor of the likeli-

0090-4295/05/$30.00doi:10.1016/j.urology.2004.12.013

htiliccawrtc

waws(mdRaw

ooRfR

f6iTwstwtd

otp(vvn

d(pv

sssam

(1Kmp2(

smlmt((auc1Pr

soPgt

C

P

AP

G

E

M

S

K

U

ood of metastatic progression.16–20 At our institu-ion, we have observed durable responses to RT inndividuals with unfavorable clinical and patho-ogic parameters and have observed that PSAVdentifies those individuals with isolated local re-urrence. Therefore, we prospectively selectedandidates for salvage RT on the basis of the PSAVlone. In this retrospective analysis, we examinedhether the response rates to RT after biochemical

elapse can be predicted and ultimately improvedhrough the use of PSAV as a solitary selectionriterion for RT.

MATERIAL AND METHODS

We retrospectively analyzed a total of 48 patients treatedith salvage RT alone for isolated biochemical relapse, defined

s a PSA level greater than 0.1 ng/mL, after RP. All patientsho developed a PSA level of greater than 0.1 ng/dL after

urgery were followed up with serial PSA measurements at 0time of PSA relapse), 3, and 6 months. All 48 patients whoaintained a PSA level of less than 1.0 ng/mL at 6 months after

etection of biochemical recurrence subsequently underwentT. Patients whose PSA level exceeded 1.0 ng/mL at 6 monthsfter relapse were excluded from this analysis, because theyere not offered salvage RT.All patients treated with salvage RT underwent a full course

f pelvic RT ranging in dose from 64 to 72 Gy at the discretionf the managing radiation oncologist. Patient follow-up afterT consisted of serial PSA measurements at 3-month intervals

or the first 2 years and then every 6 months. The response toT was defined as a PSA level of less than 0.1 ng/mL.The overall response to RT was evaluated at the patient’s last

ollow-up visit. Individual PSAVs were calculated for the-month interval before RT. The PSAV was defined as a changen PSA in nanograms per milliliter/change in time in months.he 0, 3, and 6-month PSA values were plotted for all patients,ith the x-axis representing PSA values and the y-axis repre-

enting the time in months. Because the rise in PSA levels inhese patients best fit a linear increase, a linear regression lineas fit to the three points. The slope of this regression line was

hen calculated with a common denominator of 1 month toepict the PSAV.Cox regression analysis was performed to determine the

verall association between all clinical variables and the timeo relapse. Patients were also stratified on the basis of variousathologic and clinical parameters treated as binary variablesGleason score of 7 or less versus greater than 7, presenceersus absence of extracapsular extension [ECE], presenceersus absence of seminal vesicle invasion, and positive versusegative surgical margins).Cox regression and Kaplan-Meier survival functions were

etermined using the Statistical Package for Social SciencesSPSS, Chicago, Ill). An analysis of the predictive ability of theathologic and clinical parameters was performed with uni-ariate analysis (JMP software, SAS Institute, Cary, NC).

RESULTS

The clinical and pathologic characteristics of thetudy population, including age, TNM pathologictage, range of preoperative PSA levels, Gleasoncore, ECE status, seminal vesicle involvement,nd margin status are summarized in Table I. The

ean time to PSA recurrence was 11.9 months t

ROLOGY 65 (5), 2005

range 3 to 54). The median follow-up after RT was6 months (mean 17.4, range 2 to 44) after RT.aplan-Meier survival analysis indicated that theedian postsalvage relapse-free interval for all 48

atients was 27 months (95% confidence interval0 to 34). The mean time to relapse was 29 months95% confidence interval 24 to 34).Of the 48 patients, 30 (62.5%) maintained a re-

ponse to RT (PSA level of less than 0.1 ng/mL) ataximal follow-up. Of the 27 men with a PSAV of

ess than the mean (0.034 ng/mL/mo), 22 (81.5%)aintained a PSA level of less than 0.1 ng/mL at

heir last follow-up visit compared with only 838.1%) of 21 men with a PSAV greater than meanP � 0.0013). Cox regression analysis using PSAVs the covariate strongly predicted the risk of fail-re after salvage RT. The Cox regression coeffi-ient for PSAV was 19.56 (chi-square � 5.86, df �, P � 0.013). This coefficient indicates that asSAV increases by 0.001 ng/mL/mo, the risk ofelapse increases by a factor of 1.02.Analysis of the increasing PSAV values demon-

trated that at a cutoff of 0.035 ng/mL/mo, the riskf relapse was double that observed at the minimalSAV. Using this value of PSAV as the cutpointave a true-positive fraction for relapse of 74.2%, arue-negative fraction for relapse of 70.6%, a posi-

TABLE I. Distribution of clinical,preoperative, and pathologic characteristicsharacteristic Total (n) % Mean

SA (ng/mL) 11.08�10 29 60.4�10 19 39.6

ge (yr) 59athologic stageT2a 2 4.2T2b 2 4.2T2c 13 27.1T3a 10 20.8T3b 4 8.3T3c 16 33T4a 1 2.1leason score�7 30 62.5�7 18 37.5

xtracapsularextension

Positive 26 54.2Negative 22 45.8argin statusPositive 28 58.3Negative 20 41.7

V invasionPositive 17 35.4Negative 31 64.6

EY: PSA � prostate-specific antigen; SV � seminal vesicle.

ive predictive fraction of 81.2%, and a negative

943

pbcs

ttmwlmptwtrv0

GittwsavEis

ms

2cpersbm

gworst(pmnlasPlsmtscw

slctibPwta

omPdPc8tPc4Msft

FfSvom

9

redictive fraction of 60.0% for predicting a dura-le response to RT. Raising or lowering the PSAVutpoint provided no significant improvement inensitivity or positive predictive fraction.Using the PSAV value of 0.035 ng/mL/mo, pa-

ients were divided into two groups (PSAV greaterhan 0.035 ng/mL/mo versus PSAV 0.035 ng/L/mo or less). Kaplan-Meier survival analysisas performed, resulting in the postsalvage re-

apse-free survival curves given in Figure 1. Theedian postsalvage relapse-free survival time for

atients who underwent RT with a PSAV of lesshan 0.035 ng/mL/mo was 28 months comparedith 16 months for patients with a PSAV greater

han 0.035 ng/mL/mo. A log-rank test of this factorevealed a significant difference between the sur-ival curves (log-rank statistic � 8.95, df � 1, P �.003).Using univariate Cox regression analysis, theleason score, ECE status, margin status, and sem-

nal vesicle invasion status were not predictive ofhe response to RT (Table II). The relationship be-ween PSAV and various pathologic parametersas also evaluated. In univariate chi-square analy-

is, no significant difference in PSAV was notedmong patients with a Gleason score of 7 or lessersus greater than 7 (P � 0.06), with or withoutCE (P � 0.71), with or without seminal vesicle

nvasion (P � 0.11), or with positive or negativeurgical margins (P � 0.88).

COMMENT

RP provides excellent cancer control for mosten with clinically localized prostate cancer. De-

IGURE 1. Kaplan-Meier survival analysis of relapse-ree survival after salvage RT for biochemical relapse.ignificant difference (P � 0.003) in relapse-free sur-ival found between men with PSAV of 0.035 ng/mL/mor less and those with PSAV greater than 0.035 ng/mL/o.

pite earlier detection of disease, approximately t

44

0% to 35% of RP-treated men will experience bio-hemical relapse by 10 years of follow-up.1–4 Ap-roximately 30% of these men will develop clinicalvidence of metastatic disease within 10 years ofelapse. Several studies have attempted to identifypecific clinical and pathologic characteristics thatest predict for isolated local recurrence amongen with biochemical relapse.1,3,4,16,21–24

Pound et al.19 reported that 33% of men under-oing RP will have a detectable PSA elevationithin 10 years after surgery. In that study, the riskf developing metastatic disease after biochemicalecurrence was increased in those with a Gleasoncore greater than 7, preoperative PSA level greaterhan 20 ng/mL, locally advanced pathologic stagepT3 disease), and shortened (less than 1 year)ostoperative interval to recurrence. That theseen ultimately developed metastatic disease doesot preclude the possibility that they had isolated

ocal disease at PSA recurrence. Cadeddu et al.5lso demonstrated that those men with a Gleasoncore of 8 or greater, seminal vesicle invasion, orSA recurrence within 1 year of postoperative fol-

ow-up rarely demonstrated a durable response toalvage RT. In their study, the median pretreat-ent PSA level was 2.8 ng/mL.5 Although earlier

reatment may have improved the durable re-ponse rates, the investigators did not attempt toorrelate the disease-free survival (DFS) responseith the pretreatment PSA level.Several investigators have evaluated the relation-

hip of the PSA doubling time (PSA-DT) with theikelihood of metastatic relapse.16–20 Trapasso andolleagues16 demonstrated that a PSA-DT of lesshan 4.3 months was associated with a significantlyncreased risk of metastatic relapse (proven byone scan or other radiologic imaging) and aSA-DT greater than 11.7 months was associatedith local recurrence. Partin et al.20 also reported

hat a rapid PSA recurrence and short PSA-DT isssociated with systemic metastasis.In 1997, Patel and associates18 evaluated 77 cases

f PSA recurrence. At a mean follow-up of 34onths, they found that 80% of the patients with aSA-DT greater than 6 months remained clinicallyisease free but only 64% of patients with aSA-DT of less than 6 months did so. Pruthi et al.22

orrelated PSA-DT with pathologic features in0 patients with biochemical recurrence after pros-atectomy. Multivariate analysis revealed thatSA-DT was significantly related to the extent ofapsular involvement, percentage of Gleason scoreor 5, lymph node involvement, and patient age.ost recently, D’Amico and colleagues25 demon-

trated that PSA-DT at relapse strongly predictedor the risk of death from prostate cancer. Al-hough all the above-mentioned studies support

he hypothesis that a rapid PSA-DT portends a

UROLOGY 65 (5), 2005

ppPs

rsbliesfPpmt�5A2m

ptcidt1tpcomsRir

imdsl

p

grtvatsfwools

oi1ttocWhtc

rrstkcbllPRPvo

P

GSME

K

U

oor prognosis, none attempted to identify an ap-licable treatment algorithm based on PSAV orSA-DT or correlated these findings with the re-ponse to RT.Other investigators have found a significant cor-

elation between the presalvage PSA level and re-ponse to RT. In men undergoing salvage RT foriochemical relapse, a pretreatment PSA level ofess than 0.6 ng/mL was associated with significantmprovement in 5-year DFS compared with all oth-rs (72% versus 32%, P � 0.04).15 Schild et al.26

howed a marked difference in DFS at 3 years inavor of those undergoing RT with a pretreatmentSA level of 1.1 ng/mL or less (P � 0.02). An im-roved DFS rate was reported by Song et al.13 foren undergoing salvage RT with a PSA level less

han 1 ng/mL (60% versus 22% 4-year DFS rate, P0.01). Vicini and associates7 reported an overall

-year DFS rate of only 16% after salvage RT.mong men with a pre-RT PSA level of less than.0 ng/mL, however, the response was approxi-ately 80%.7Our series is the first to use PSA to select men

rospectively for salvage RT using PSAV alone. Al-hough the PSAV is distinct from the PSA-DT, itan be considered a surrogate for the PSA-DT, as its intuitive that a rapid velocity will result in a shortoubling time. We empirically treated those pa-ients with a slowly rising PSA level (less than.0 ng/mL after 6 months of follow-up) and nega-ive metastatic evaluation. The preoperative andathologic characteristics did not influence our de-ision of whether to proceed with therapy. Ourverall response rate of 62.5% exceeded that ofost historical series. Of note, in the only other

eries correlating PSAV and the response to salvageT, a similar 62.9% response rate after salvage RT

n men with a PSA-DT greater than 12 months waseported.14

The results of our study suggest that PSAV alonedentifies a subset of individuals who will ulti-

ately benefit from salvage RT. The PSAV pre-icted the response to salvage RT failure. The re-ponse rate was 81.5% for those men with a PSAVess than the mean (0.034 ng/mL/mo).

This study failed to demonstrate a significant

TABLE II. Univariate Cox regression

arameterParameterEstimate

leason score 0.290V invasion 0.265argin status 0.116

xtracapsular extension �0.500

EY: SE � standard error of the mean; SV � seminal vesicle.

redictive value for seminal vesical invasion, sur- 4

ROLOGY 65 (5), 2005

ical margin status, or Gleason score to predict theesponse to RT. Thus, we believe that the PSAV ishe best single parameter for selecting men for sal-age RT. It is possible that with longer follow-up,dverse pathologic features may ultimately predicthe response. Although our relatively high re-ponse rate overall may be a byproduct of the shortollow-up, a subset analysis of those individualsith a minimal follow-up of 2 years revealed anverall response rate of 61.1% (11 of 18) and a ratef 71.4% (10 of 14) for those patients with a PSAVess than the mean. As such, we anticipate our re-ults will persist with longer follow-up.The results of our study support those of previ-

us reports that good response rates are observedn men with a pretreatment PSA level of less than.0 ng/mL. We also demonstrated that the PSAV ishe best predictor of a response to salvage RT. Al-hough our overall follow-up was relatively short,ur data for individuals with longer follow-up wereonsistent with the outcomes for the whole group.

ith longer follow-up and an expanded series, weope to identify the optimal PSAV for selectinghose candidates in whom salvage RT should beonsidered for biochemical relapse after RP.

CONCLUSIONS

The PSAV strongly predicts for the likelihood ofesponse to salvage RT in men with biochemicalelapse after RP. Men with a lower PSAV have aignificantly greater likelihood of response thanhose with a higher PSAV. This study, to ournowledge, represents the first prospective appli-ation of a PSAV-based treatment algorithm foriochemical recurrence after RP. Additional fol-ow-up of our patients and the incorporation of aarger patient sample size will allow us to refine ourSAV cutoff and determine the impact of salvageT on patient survival. The use of a well-definedSAV cutoff as the sole selection criterion for sal-age RT after RP may improve the historically lowverall response rates.

REFERENCES1. Carroll P: Rising PSA after a radical treatment. Eur Urol

alysis for all pathologic parameters

E Wald df P Value

262 1.222 1 0.269518 0.262 1 0.609529 0.048 1 0.826515 0.943 1 0.331

an

S

0.0.0.0.

0: 9–16, 2001.

945

rp

caH

yi1

rl

pc1

ct

tf

ap

rc1

aa2

rc5

rt

rt

fg

ds1

tfM

aps

hp

sts

ctl

apt

pr

cp2

pt1

rp

9

2. Catalona WJ, and Smith DS: 5-year tumor recurrenceates after anatomical radical retropubic prostatectomy forrostate cancer. J Urol 152: 1837–1842, 1994.

3. Han M, Partin AW, Pound CR, et al: Long-term bio-hemical disease-free and cancer-specific survival followingnatomic radical retropubic prostatectomy: the 15-year Johnsopkins experience. Urol Clin North Am 28: 555–565, 2001.

4. Zincke H, Oesterling JE, Blute ML, et al: Long-term (15ears) results after radical prostatectomy for clinically local-zed (stage T2c or lower) prostate cancer. J Urol 152: 1850–857, 1994.

5. Cadeddu JA, Partin AW, DeWeese TL, et al: Long-termesults of radiation therapy for prostate cancer recurrence fol-owing radical prostatectomy. J Urol 159: 173–178, 1998.

6. Garg MK, Tekyi-Mensah S, Bolton S, et al: Impact ofostprostatectomy prostate-specific antigen nadir on out-omes following salvage radiotherapy. Urology 51: 998–1002,998.

7. Vicini FA, Ziaja EL, Kestin LL, et al: Treatment out-ome with adjuvant and salvage irradiation after radical pros-atectomy for prostate cancer. Urology 54: 111–117, 1999.

8. Nudell DM, Grossfeld GD, Weinberg VK, et al: Radio-herapy after radical prostatectomy: treatment outcomes andailure patterns. Urology 54: 1049–1057, 1999.

9. Pisansky TM, Kozelsky TF, Myers RP, et al: Radiother-py for isolated serum prostate specific antigen elevation afterrostatectomy for prostate cancer. J Urol 163: 845–850, 2000.10. Leventis AK, Shariat SF, Kattan MW, et al: Prediction of

esponse to salvage radiation therapy in patients with prostateancer recurrence after radical prostatectomy. J Clin Oncol9: 1030–1039, 2001.11. Do LV, Do TM, Smith R, et al: Postoperative radiother-

py for carcinoma of the prostate: impact on both local controlnd distant disease-free survival. Am J Clin Oncol 25: 1–8,002.12. Chawla AK, Thakral HK, Zietman AL, et al: Salvage

adiotherapy after radical prostatectomy for prostate adeno-arcinoma: analysis of efficacy and prognostic factors. Urology9: 726–731, 2002.13. Song DY, Thompson TL, Ramakrishnan V, et al: Salvage

adiotherapy for rising or persistent PSA after radical prosta-ectomy. Urology 60: 281–287, 2002.

14. Peyromaure M, Allouch M, Eschwege F, et al: Salvageadiotherapy for biochemical recurrence after radical prosta-

ectomy: a study of 62 patients. Urology 62: 503–507, 2003. J

46

15. MacDonald OK, Schild SE, Vora SA, et al: Radiotherapyor men with isolated increase in serum prostate specific anti-en after radical prostatectomy. J Urol 170: 1833–1837, 2003.

16. Trapasso JG, deKernion JB, Smith RB, et al: The inci-ence and significance of detectable levels of serum prostatepecific antigen after radical prostatectomy. J Urol 152: 1821–825, 1994.17. Roberts SG, Blute ML, Bergstralh EJ, et al: PSA doubling

ime as a predictor of clinical progression after biochemicalailure following radical prostatectomy for prostate cancer.

ayo Clin Proc 76: 576–581, 2001.18. Patel A, Dorey F, Franklin J, et al: Recurrence patterns

fter radical retropubic prostatectomy: clinical usefulness ofrostate specific antigen doubling times and log slope prostatepecific antigen. J Urol 158: 1441–1445, 1997.

19. Pound CR, Partin AW, Eisenberger MA, et al: Naturalistory of progression after PSA elevation following radicalrostatectomy. JAMA 281: 1591–1597, 1999.20. Partin AW, Pearson JD, Landis PK, et al: Evaluation of

erum prostate-specific antigen velocity after radical prosta-ectomy to distinguish local recurrence from distant metasta-es. Urology 43: 649–659, 1994.

21. Kupelian P, Katcher J, Levin H, et al: Correlation oflinical and pathologic factors with rising prostate-specific an-igen profiles after radical prostatectomy alone for clinicallyocalized prostate cancer. Urology 48: 249–260, 1996.

22. Pruthi RS, Johnstone I, Tu IP, et al: Prostate-specificntigen doubling times in patients who have failed radicalrostatectomy: correlation with histologic characteristics ofhe primary cancer. Urology 49: 737–742, 1997.

23. Koch MO, Foster RS, Bell B, et al: Characterization andredictors of prostate specific antigen progression rates afteradical retropubic prostatectomy. J Urol 164: 749–753, 2000.

24. Pound CR, Partin AW, Epstein JI, et al: Prostate-spe-ific antigen after anatomic radical retropubic prostatectomy:atterns of recurrence and cancer control. Urol Clin North Am4: 395–406, 1997.25. D’Amico AV, Moul JW, Carroll PR, et al: Surrogate end

oint for prostate cancer-specific mortality after radical pros-atectomy or radiation therapy. J Natl Cancer Inst 95: 1376–383, 2003.26. Schild SE, Buskirk SJ, Wong WW, et al: The use of

adiotherapy for patients with isolated elevation of serumrostate specific antigen following radical prostatectomy.

Urol 156: 1725–1729, 1996.

UROLOGY 65 (5), 2005