- Email: [email protected]
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN IN THE FOLLOWUP OF 422 PATIENTS AFTER RADICAL PROSTATECTOMY
0022-5347/99/1614-1206/0
Vol. 161, 1206-1211, April 1999 Printed in U.S.A.
THE JOVRVAL OF UROLOGY Cop-mght 0 1999 by AMERICA\ U H O ~ I C A.SSOCUTIO.U, AL Isc
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN IN THE FOLLOWUP OF 422 PATIENTS AFTER RADICAL PROSTATECTOMY ALEXANDER HAESE, EDITH HULAND, MARKUS GRAEFEN, PETER HAMMERER, JOACHIM NOLDUS AND HARTWIG HULAND From the Department of Urology, Unillersity Clinic Eppendorf, Hamburg, Germany
ABSTRACT
Purpose: We validated our ultrasensitive prostate specific antigen (PSA) assay based on lyophilization and 4-fold concentration of patient sera with the clinical long-term followup and according to histopathological characteristics of 422 patients treated with radical retropubic prostatectomy for prostate cancer. Materials and Methods: Each serum sample was divided into 2 aliquots for standard and 4-fold concentrated (ultrasensitive) detection. Samples were analyzed by the same unmodified DPCImmulite* PSA assay. Biochemical relapse was defined as a n increase of at least 0.10 ng./ml. in native serum (equivalent to 0.025 ng./ml. in concentrated serum). Mean followup was 449 days (range 29 to 2,057 ). Kaplan-Meier analysis of standard and ultrasensitive detection results was done, and findings were correlated with pathological stage, Gleason grade, total cancer volume, Gleason grade 4 cancer volume and margin status. Significance of earlier detection in ultrasensitive versus standard detection w a s calculated with the log rank (Mantel-Cox)test with p c0.05 considered significant. Results: Of 442 patients 88 (20.8%)experienced biochemical recurrence. Of this cohort 28 (31%) demonstrated early failure on the ultrasensitive assay which was later confirmed on the standard assay, 37 (42%)had failure simultaneously on both assays and 23 (26%)had failure on the ultrasensitive but remained disease-free on the standard assay. Average time for ultrasensitive assay detection of recurrence was 288 days (standard 555). Kaplan-Meier analysis revealed significant advantages in earlier detection of recurrence with the ultrasensitive assay, and close correlation with pathological stage, Gleason grade, margin status and Gleason grade 4 cancer volume. Time advantages of ultrasensitive versus standard detection were greater for advanced cancers (pTSa/b or greater, Gleason 3 + 4 or greater) than for small, low grade tumors. All patients who had positive results on the standard assay had a previous (28) or simultaneous (37) positive ultrasensitive result. With standard detection 25% of all relapses were evident within the first year of surgery and with ultrasensitive detection the percentage increased to 85.7%. On both assays 334 patients remained free of biochemical recurrence. Conclusions: Our ultrasensitive PSA assay is useful for early detection of biochemical relapse after radical retropubic prostatectomy. It not only provides the same accuracy as conventional PSA assays but also offers the advantage of detecting recurrence about 300 days earlier. Thus, long-term results of radical retropubic prostatectomy series can be calculated sooner. The clinical impact of this assay will be obvious once curative treatment options are available if applied a t the earliest time of evident tumor recurrence. KEY WORDS:prostate-specific antigen, prostatic neoplasms, prostatectomy About 100,000 men undergo radical prostatectomy each year in the United States for clinically localized prostate cancer.' About 30 to 50% of these patients ultimately have biochemical evidence of failure measured by prostate specific antigen (PSA), an organ specific biochemical marker of recurrent disease.Z-" When first generation assays with a PSA detection limit of 0.2 to 0.6 ng./ml. were used, PSA recurrence usually preceded clinical recurrence by 2 to 5 A newer generation of assays, t h e so-called ultrasensitive assays, lowered the PSA detection limit by a factor of about 5 to 10, permitting earlier Work on assays has focused on increasing this sensitivity by lowering PSA detection limits.'4-") An alternative approach is the manipulation Accepted for publication November 13, 1998. hesented at annual meet,ng of ~ ~Association, ~ San Diego. California, May 30-June 4, 1998. * Diagnostic Product Corp.. Los Angeles. California.
of the serum sample. We recently described an ultrasensitive method to detect PSA at followup after radical prostatectomy based on concentration of serum samples.21.'* The validity of PSA values below the detection limit of a particular assay can be tested against background with t h e same (unmodified) assay. True PSA values below t h e detection limit should concentrate above the detection limit and, thus, can be identified, whereas background noise theoretically should not concentrate. By identifying true positive PSA values below the detection limit we can detect PSA recurrence earlier. We validate the clinical applicability of serum concentration in a large series of patients observed for up to 5 years to determine whether this technique would result in earlier detection of recurrent prostate cancer after radical prostatectomy. For clinical validity we correlated of ultrasensitive l ~ ~ , t h e results ~ ~ ~ and standard detection to histopathological features of prostate specimens.
1206
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN MATERIALS AND METHODS
Patient selection and evaluation. We prospectively analyzed postoperative sera from 422 consecutive patients who underwent pelvic lymphadenectomy and radical retropubic prostatectomy for clinically localized prostatic carcinoma. All lymph nodes were negative for prostate cancer metastasis on pelvic lymph node dissection. Patients who received preoperative hormonal therapy were excluded from study. Patients were seen during regularly scheduled visits at our outpatient department, and followup began at 8 to 10 weeks after surgery. Patients typically were seen after 6, 12, 18 and 24 months, and then at 1-year intervals. A total of 1,046 serum samples were analyzed and the number of followup samples ranged from l(176 patients) to 12 (1). Ofthe 422 patients 246 (58.3%) had at least 2 consecutive measurements during followup. Followup ranged from 29 to 2,057 days or 5.6 years (mean 449, median 327 days). Length of followup was less than 1 year in 224 patients (53.1%),more than 1 and up to 3 years in 154 (36.5%),more than 3 and up to 5 years in 40 (9.5%),and more than 5 years in 4. Principles of lyophilization and concentration. Consecutive serum samples from each patient were drawn at each visit before digital rectal examination and transrectal endosonography. Patient sera were brought to our laboratory and divided into 2 aliquots of 1 ml. each (samples 1 and 2). Both samples were frozen a t -8OC. Sample a was left untreated and sample 2 was lyophilized, which resulted in a 10% dry weight powder of the original sample. A standardized quantity of 150 p1. distilled water was added, which resulted in a 250 pl. 4-fold concentrated serum sample. PSA detection. For PSA analysis we used an Immulite PSA assay, which is a solid phase, 2-site chemiluminescent immunometric assay with a polyclonal-monoclonal antibody combination. Analytical PSA sensitivity is 0.04 ng./ml., whereas clinical sensitivity is 0.10 ng./ml.** The untreated (native) and concentrated (ultrasensitive) sera were analyzed by the conventional assay with standard reagent kits and standard master curves. Native and concentrated samples were measured in triplicate. Study design. We continuously monitored PSA after radical prostatectomy in each patient, and compared the results of native and ultrasensitive detection of each sample. Results of both assays showing no PSA recurrence (defined as an increase from less than 0.10 to at least 0.10 ng./ml.) were noted on a specific day as negative or censored. Ultrasensitive value exceeding our definition of positivity (0.10 ng./ml.) but PSA less than 0.10 ng./ml. for native detection was noted as uncensored and censored for ultrasensitive and native samples, respectively. If on a later measurement both samples exceeded 0.10 ng./ml., the native sample was termed uncensored as well. The time difference between the appearance of an uncensored value on the concentrated sample in comparison to the native sample was the time advantage that could be achieved. Recurrence was correlated with various characteristics, including pathological stage, Gleason grade, margin status, cancer volume and Gleason grade 4 cancer volume. For all patients results for native and concentrated sera and significance tests and analyses of PSA-free survival were compared. Histological characterization of the prostate. The prostate was prepared according t o the Stanford protocol. It was inked over the entire surface, fixed in formalin for at least 24 hours and processed with a 3 mm. step section technique. The Gleason system was used for histological grading and the TNM classification was used for staging. Cancer volume was calculated with a computer assisted volumetric program developed at our department of pathology. Additional histological characterization included margin status, total cancer volume and Gleason grade 4 cancer volume. Statistical analysis. To calculate the significance of time
1207
differences between the appearance of uncensored values in standard versus ultrasensitive detection we performed a Kaplan-Meier analysis of PSA-free survival for native and concentrated samples separately. The 2 curves were superimposed for maximal information gain. Significance was calconsidered culated with the Mantel-Cox test with p ~ 0 . 0 5 significant. RESULTS
Of 442 patients 88 (20.86) experienced biochemical recurrence. Of this cohort 28 (31%)demonstrated early failure on the concentrated assay which was later confirmed on the standard assay, 37 (42%)had failure simultaneously on both assays and 23 (26%) had failure on the concentrated but remained disease-free on the standard assay. This study involved a dynamic process and so the confirmation of biochemical failure, defined as an increase in PSA to at least 0.10 ng./ml. in native samples, particularly in the last group of patients, must be awaited. Therefore, evaluation of our ultrasensitive PSA analysis was based on patients who had a positive result first on concentrated and later on native samples. Of 88 patients 28 fulfilled these criteria (table 1).Mean time to evidence of relapse was 555.2 days (range 129 to 1,389), for these patients, whereas the ultrasensitive assay detected PSA relapse a mean of 288 days (range 47 to 1,130) after radical prostatectomy. Thus, a mean lead time of 267 days (range 66 to 650) was possible, which has a great potential benefit for patients. PSA relapse was detected on native samples within the first year after radical prostatectomy in 7 of 28 patients (25%).The ultrasensitive assay detected PSA relapse within the first year in 24 of the 28 patients (85.7%).On both assays 334 patients remained free of biochemical recurrence. The relative frequency of failure compared with pathological stage, Gleason score, total cancer volume, Gleason grade 4 cancer volume and margin status is shown in table 2 for standard and ultrasensitive detection, and for the 23 patients who had positive results with ultrasensitive but negative results with standard detection.
TABLE1 Summary of 28 patients uvth positiLv resuits rnrtrallv on ultrasensitive and later on standard PSA assajb First Pos Result tdaj)
Pt No
~
Native
1 2 3 4 5 6 7
557 368 1,010 408 1.192 378 556 1,395 373 3 75 572 196* 476 581 728 656 720
Concentrated
- -
Dtflerence
249 193* 1 75 573 437 209* 199 360 832 193' 185 365' 191 59s 8 800 176 9 197* 160' 215 10 215 357* 11 126 12 70* 211 13 265* 407 174' 14 650 15 78' 327 16 323* 488 17 232' 367 303 18 64* 217* 126 91' 19 96 206302* 20 259 1,389 1.130 21 213 287* 72' 22 140 202* 34'" 23 66 129* 63. 24 282 595 313' 25 210 535 325* 26 364 242* 606 27 195 47* 28 242' __ Mean (range) 555 (129-1.389) 288 (47-1.130) 267 (66-650) * Biochemical failure 1 year or less aRer surgery on standard assay in 7 of 28 patients (254%)and on ultrasensitive assay in 24 of 28 patients (85.7%). 308'
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN
1208
TABLE2. Comparison of histopathological characteristics and PSA relapse on ultrasensitiue and natiue assays No. Ultrasensitive Assay Pos. Only ( 7 6 ) No. Both Assays Pos. (B) Total No. Pts. Pathological stage: pT2a-c
2 (1.11 25 (20.6) 33 (47.81 5 (50) 65
3 (1.6) 12 (9.9) 8 111.59)
136 163 62 13
1 (0.71 23t14.1) 31 (50.0) 10 (76.9) 65
0 10
(6.1) 11 (17.71 -2 (15.41 23
69 297
29 (42.2) 33 (11.0) 62
4 16 20
0 3 (5.4) 13 f 15.9) 12 (20.7) 24 (44.41 52
0
185 121 69 10
pT3ah
pT3c PT4 Totals Gleason score: 3 + 3 or Less
314 4+3 4 + 4 or Greater Totals
Margin status: Pos. Neg. Totals Tumor vol. (cc): Less than 0.5
22 56 82 58 54
0.5-2 2-5 5-10 Greater than 10 Totals
- -
23
(5.8) (5.41
0 5 (6.1) 10 (17.21 2 (3.7) 17
Gleason grade 4 tumor vol. (ccl:
0
117 128 73 30 40
Less than 0.5 0.5-2
2-5 Greater than 5
0
4 (3.1) 13 (17.8) 12 (40.01 23 (57.51 52
- _ Totals _ _ _ ~ Kaplan-Meier curves for PSA-free survival are shown in figure 1 for all patients (p = 0.03). These results were stratified according to histopathological features of specimens. The respective Kaplan-Meier curves in figure 2 show significance according to the log rank (Mantel-Cox) test. Comparison of standard and ultrasensitive detection revealed significant advantages in favor of the latter for pathological stage margin status (p = 0.03) (p = 0.016), Gleason grade (0.007), and Gleason grade 4 cancer volume ( p = 0.027). Since there was no recurrence in patients with less than 0.5 cc total cancer volume, a p value could not be calculated for this parameter. To determine the time benefit of ultrasensitive detection of PSA we created a Kaplan-Meier curve that included all patients who had positive results for native and concentrated sera. Patients with positive results on ultrasensitive but negative results on standard detection were excluded from analysis. Figure 3 shows the time advantage expressed as the area between curves (p = 0.04). DISCUSSION
The first generation PSA assays used clinically had a detection limit of 0.2 to 0.6 ng./ml. However, for followup of
1 ,
-
..
.e Cum SuMval ( a d ) Cum Survival (sup)
u
.2
'
-Chi-square 4,655-
CF-
P-Value.
1 ,
,0310
.-
0-., 0
500
1000
1500
2000
Time
FIG. 1. Kaplan-Meier curve shows PSA-free survival in standard (std)and ultrasensitive (sup)detection for all patients. Cum., Cumu-
lative.
0 1 (0.78) 2 (2.71 7 (23.3) 10 (25) 20
radical prostatectomy patients a lower detection limit of 0.2 to 0.6 ng./ml. proved insufficient for the minute changes in serum PSA that indicated recurrence of prostate cancer. Therefore, various efforts have been made to improve the sensitivity of PSA assays to reduce the time between radical prostatectomy and evidence of biochemical relapse. In 1992 Graves et a1 reported a successful method which increased the analytical sensitivity of the standard assay from a working range of 0.3 to 50 to 0.1 to 1.2 ng./ml. with a modification of a radioimmunoassay.20 PSA values in 2 patients with recurrent prostatic cancer increased above the 0.1 ng./ml. detection limit of the modified assay 175 and 581 days before reaching the 0.3 ng./ml. detection limit of the standard assay, which demonstrates a significant time advantage of potential benefit to patients with respect to beginning adjuvant therapy earlier. In the screening of 187 post-radical prostatectomy patients for whom the standard assay revealed no evidence of cancer the ultrasensitive assay described by Graves et a120 was applied by Stamey et all4 and detected evidence of residual cancer (PSA at least 0.1 ng./ml.) in 21 patients (11.2%).In the same study simultaneous analysis with both assays was performed on sera of 22 patients with proved recurrence. The ultrasensitive assay (sensitive to PSA at at least 0.1 ng./ml.) detected PSA recurrence much earlier in all 22 patients than the standard assay (sensitive to PSA at least 0.3 ng./ml.) by a median and mean of 202 and 310 days, respectively. The technical backbone of the ultrasensitive assay was a longterm incubation of samples and the creation of a standard curve which was adjusted to the extremely low PSA values in post-radical prostatectomy sera. This modification precluded the use of the assay in the followup of patients with low serum PSA concentrations. Yu and Diamandis reported an ultrasensitive method for measuring PSA in serum, including a monoclonal-polyclonal antibody combination and alkaline phosphatase labeled ~treptavidin.'~ The main characteristic was sensitivity as it could detect PSA as low as 0.002 nglml. Based on a monoclonalmonoclonal combination and 1-step incubation as a further improvement Ferguson et a1 demonstrated a lower limit of
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN
1209
Cum Survival (llsld)
cum survival (1lMd) Cum Survival (llarp) Cum Survival (ZIsldJ
Cum. Survival (Ilsup) Cum. Survival (ZIed)
Cum Survival (2/supJ
Cum. Survival (Zsup) Cum. Survival (Ysld)
g
Cum. Survival (Ysup)
0
1000
500
1500
,4
o
cum Sumval (3mj
-1
Cum Survival (Yaup)
i
Cum Survival (41816)
Cum survnval (4lsld)
Cum. Survival (Usup)
cum SuMval (4lS"P)
0
2000
500
Time
1000
1500
2000
TlmR
D Cum. Survival (yee/¶d) Cum. Sumval (yedsup)
g 4
1iirr' Chi-Sqare
0
Cum Survival (noled)
L
-_
4,368
Cum. Survival (ndsup)
F P-Value
,0366
!
1
_?__i
500
0
I000
1500
2000
Time
E
-
'
~-
14
L-
Cum. Survival (zerolsld) Cum. survival Izom/sup) Cum Survival (
1
1 t
i o
500
iooo
1500
Cum survival (0.5-2/sup) Cum. Survival (,2-5/sId) Cum Survival (>Z-SIsup) Cum. Survival (>wnd) cum. survivai (>%sup)
zoo0
Time
FIG. 2. Kaplan-Meier curves. A, PSA-free survival in standard (std) and ultrasensitive (sup)detection for pathological stages 2a to c ( I ) , 3ah (2),3c ( 3 )and 4 ( 4 ) . B,PSA-free survival in standard and ultrasensitive detection for Gleason scores 3 + 3 or less ( I ) , 3 + 4 ( Z ) , 4 + 3 (3) and 4 + 4 or greater ( 4 ) . C , PSA-free survival in standard and ultrasensitive detection for positive margin (yes) and negative margin. D , PSA-free survival in native and ultrasensitive detection for total tumor volume in 5 groups. Because there was no recurrence in group with less than 5 cc tumor volume log rank (Mantel-Cox)test could be calculated. E , PSA-free survival in standard and ultrasensitive detection for Gleason grade 4 cancer volume in 5 volume groups. Cum., Cumulative.
'
Cum. Survival (std) Cum. Survival (sup)
t 0
400
800
1200
1600
Time
FIG. 3. Kaplan-Meier curve shows PSA-free survival in standard (std) and ultrasensitive (sup)detection for 65 patients with ultrasen-
sitive and standard evidence of biochemical failure. Area between curves represents time benefit achieved by lyophilization concentration. C u m . , Cumulative.
detection of better than 0.001 ngfml.15 Comparison with a commercially available third generation assay revealed substantial PSA changes at concentrations undetectable with conventional assays. A retrospective study with the ultrasensitive method of Ferguson et a l l 6 was performed by Yu et al. in 148
patients with a postoperative PSA of less than 0.1 ngfml. according to a conventional PSA assay.**Increases in postoperative serum PSA at levels of 0.001 to 0.1 ngfml. after radical prostatectomy were associated with clinicopathologicalfeatures of poor prognosis, and they concluded that monitoring patients with a highly sensitive PSA assay (detection limit 0.001 ngfml.) was a simple and effective method of detecting clinically important biochemical relapse soon after radical prostatetomy. This was one of the first studies to correlate biochemical failure on extremely low concentrations of serum PSA with histopathological features of prostate cancer. Positive margins (p <0.01), increasing tumor volumes ( p <0.01), higher preoperative PSA (p = 0.03) and extracapsular penetration (p = 0.02) were significantly associated with recurrence, and recurrence could be detected by measuring PSA at concentrations not detected by standard assays. Our approach to increase the sensitivity of PSA assays is different. Rather than modifying the assay specifically, for example by increasing the incubation time or developing a master curve adjusted to low levels of PSA described by Graves et a1,20 improving antibodies or introducing monoclonal-monoclonal combinations and so forth, we focused on the serum sample. This approach has the advantage of leaving well established assay designs untouched, permits
1210
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN
the application of our technique in each laboratory without large investments, and permits post-radical prostatectomy samples to be analyzed with the normal standard assay and even in the same test run with routine samples of preoperative sera. The serum concentration techniques were first described by us.22 In another study Pruthi et al retrospectively analyzed sera by 2 concentration modes of lyophilization and ultrafiltration.21 Sera from women, cystoprostaktomy patients, those cured by radical prostatectomy, those with benign prostatic hyperplasia and those whose prostate cancer was treated with radiation or hormonal therapy were used as controls. The primary study group consisted of 31 patients who had recurrent disease after radical prostatectomy and whose initial postoperative PSA decreased to undetectable levels (less than 0.07 ngJml.) and later was positive (greater than 0.07 ngJml.) according to the TOSOH AIA* 600 immunoassay using the ultrasensitive mode. Sera concentrated by lyophilizationand filtration demonstrated PSA recurrence significantly earlier than unconcentrated sera in 29 (94%)and in 28 (90%)of the 31 patients, respectively. Mean time advantage was 362 days for 29 patients and 383 for 28, which correlated well with our results of 308 to 366 days.2" Sera from women, cystoprostatectomy patients and those cured by radical prostatedomy failed to concentrate, yielding a test specificity of 100%as in our study.22 Pruthi et al concluded that serum concentration, irrespective of the method used, is a specific and sensitive technique which provides a significant lead time of about 12 months to detect cancer recurrence after radical prostatectomy. More important, the combination of serum concentration with a more advanced assay (the AIA) gave an additional lead time of 9 months at a 0.07 ngJml. residual cancer detection limit, which yielded a total benefit of about 2 years compared with first generation assays which had a 0.2 ngJml. residual cancer detection limit (Hybritech Tandem-R? assay). In 1995 Yu et al reported on ultrasensitive PSA detection (0.02 ngJml. lower limit) to monitor PSA changes in 10 patients in whom prostatectomy failed.25Prostatic cancer recurrence could be diagnosed a mean of 420 or 883 days earlier with the ultrasensitive assay than with assays that had a 0.1 or 0.3 ngJml. detection limit, respectively. They suggested that more sensitive PSA assays might be able to reduce the relapsefree periods further in about 50% of the prostatic cancer patients who ultimately have relapse. Another study comparing a first generation assay (Tandem-Et Singlepint PSA assay) with improved successors (IMxS and Tandem-E Multipoint PSA assays) was done by van Iersel et al in 1996.23 Compared with the lead time obtained with the Tandem-E Singlepoint assay, 12 of 137 patients with biochemical recurrence had a median gain in lead time of 327 days (range 60 to 627) with the IMx and 369 days (range 60 to 639) with the Tandem-E Multipoint assays. It is by rational interpretation and continuation of these studies that further improvement of future assay generations will decrease the time needed to detect recurrence of prostate cancer. However, because our method is independent of the assay itself, that need not restrict the use of serum concentration. When we study the time differences in detection between standard and ultrasensitive methods, a striking feature is that not all patients seem to benefit from ultrasensitive detection. A relatively large number of patients (37 of 88) had positive results with standard and ultrasensitive detection simultaneously despite previous negative results with both assays. In these patients there might have been a time at which the ultrasensitive assay could have detected PSA and the standard assay could not but this time was missed because of the limitations of blood sampling intervals during followup. We are aware that some patients had a time advantage with ultrasensitive detection that might well have been smaller if the sampling intervals had been shorter.
* TOSOH Medics, Foster City, California.
t H britech, Inc., San Diego. California. $ Albott Laboratories, Abbott Park, Illinois.
The relatively short mean followup of our study in our opinion does not affect our results because individual length of followup is essential. If a patient has recurrent disease relatively soon after radical prostatectomy, that is within 1 year after surgery as in 24 of 28 patients in our study, the need for longer followup is less important. On the other hand, we have patients with followup greater than 3 and 5 years, and most are free of disease. We did not evaluate the outcome of radical prostatectomy but the ability of 2 assay types to detect PSA indicative of disease recurrence. Thus, the comparison of the time differences at which recurrence is detected by the assays is important, rather than the absolute postoperative followup length before recurrence. The correlation of pathological features of the prostate specimen and results of standard and ultrasensitive detection demonstrate that, irrespective of the feature examined (for example pathological stage), the time to detect a recurrence was significantly earlier for ultrasensitive versus standard detection (fig. 2). As described by others advanced pathological stage, predominance or even presence of Gleason grades 4 and 5, and so forth in the prostate are typically associated with PSA levels that can be detected earlier using assays with increased s e n s i t i ~ i t y . ~ ~ . ~ ~ In our cohort of patients who ultimately had radical prostatectomy failure as proved by an increase in PSA from undetectable to detectable levels initially on ultrasensitive and subsequently on standard detection (281, the PSA levels were continuously increasing to levels higher than 0.10 ng./ ml. Thus, we assume it is correct to classify those cases as radical prostatectomy failure. Another 23 patients in the study group had biochemical failure on ultrasensitive detection and the increase in PSA to detectable levels on standard measurements is awaited. However, we expect at least a large number of these patients to have failure at some time, in particular those with unfavorable pathological conditions. As we have shown in table 2 the patients who have already had failure compared to those with no detectable PSA on standard measurement have a comparable distribution of pathological features. As noted previously we observed a dynamic process, and so it is not surprising that 23 patients have no proved recurrence yet. Due t o shorter followup the reconfirmation of failure has not occurred to date. The potential extraprostatic source of PSA (periurethral glands) seemed to be of no relevance in patients who had failure according to PSA detection. If PSA from those sources had mimicked PSA failure, it should have been detectable on the first postoperative serum sample. Also, in this case the level of PSA should be relatively stable at low concentrations rather than continuously increasing to levels indicative of recurrence which occurred in our study. The clinical relevance of ultrasensitive detection of PSA can be explained by its impact on the accurate assessment of the outcome after radical prostatectomy and the treatment of patients. Patients who are at risk for failure due to unfavorable pathological conditions should be monitored more closely, particularly if ultrasensitive detection is suggestive of failure. On the other hand, patients with a pathological condition that is less suggestive of recurrence and no sign of PSA recurrence on ultrasensitive detection can be followed less extensively. The established treatment modalities for recurrent prostate cancer, that is hormonal and radiation therapy, offer no cure. Moreover, in other malignancies therapies for recurrent disease offer advantages in terms of cure when instituted at the earliest time. The efficacy of early treatment of recurrent prostate cancer has not been investigated with long-term followup to our knowledge.24,~Ultrasensitive detection of PSA provides an improved tool to identify patients who might benefit from those therapies. In particular patients who are relatively young at the time of proved recurrence should be monitored intensively.
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN CONCLUSIONS
Ultrasensitive PSA detection i s useful for early detection of biochemical relapse after radical retropubic prostatectomy. It not only provides the s a m e accuracy as conventional PSA a s s a y s b u t also offers the a d v a n t a g e of yielding a positive r e s u l t a b o u t 300 d a y s earlier. Most biochemical failures can be detected within the first y e a r a f t e r radical prostatectomy. Sensitivity and specificity are equal t o those of standard PSA assays. Correlation of ultrasensitive assay findings with pathological f e a t u r e s i s equal t o that of the standard assay, which i s a sufficient rationale to apply t h i s modification at o u r laboratory. Independence of serum concentration from the a s s a y t y p e p e r m i t s use w i t h present and f u t u r e generation a s s a y s w i t h o u t questioning these improved assays. Thus, t h i s m e t h o d is a completion rather t h a n a competitor of new generation assays or a s s a y types developed b y others.19.2" An optimized PSA a s s a y m a k e s e v e n m o r e sense in combination w i t h an optimized s e r u m s a m p l e t o achieve m a x i m a l information w i t h respect t o detecting the minute changes in PSA concentrations that indicate PSA recurrence a f t e r radical prostatectomy. Long-term results of radical retropubic prostatectomy series c a n be calculated earlier. The clinical i m p a c t of t h i s a s s a y will be obvious a f t e r curative t r e a t m e n t options are available if it is used at the earliest t i m e of evident t u m o r recurrence.
REFERENCES
1. Olsson, C. A. and Goluboff, E. T.: Detection and treatment of prostate cancer: perspective of the urologist. J . Urol., 1 5 2 1695, 1994. 2. Partin, A. W., Pound, C. R., Clemens, J. Q., Epstein, J. I. and Walsh, P. C.: Serum PSA after anatomic radical prostatectomy. The Johns Hopkins' experience after 10 years. Urol. Clin. N. Amer., 2 0 713, 1993. 3. Trapsasso, J. G., deKernion, J. B., Smith, R. B. and Dorey, F.: The incidence and significance of detectable levels of serum prostate specific antigen after radical prostatectomy. J. Urol., 152: 1821, 1994. 4. Stamey, T. A., Yang, N., Hay, A. R., McNeal, J . E., Freiha, F. S. and Redwine, E. A.: Prostate-specific antigen as a serum marker for adenocarcinoma of the prostate. New Engl. J. Med., 317: 909, 1987. 5. Oesterling, J. E., Chan, D. W., Epstein, J. I., Kimball, A. W., Bmzek, D. J., Rock, R. C., Brendler, C . B. and Walsh, P. C.: Prostate-specific antigen in preoperative and postoperative evaluation of localized prostatic carcinoma treated with radial prostatectomy. J. Urol., 1 3 9 766, 1988. 6. Stamey, T. A,, Kabalin, J. N., McNeal, J. E., Johnstone, I . M., Freiha, F. S., Redwine, E. A,, Yang, N.: Prostate specific antigen in the diagnosis and treatment of adeno-carcinoma of the prostate: 11. Radical prostatectomy treated patients. J. Urol., 141: 1076, 1989. 7. Lange, P. H., Ercole, C. J., Lightner, D. J., Fraley, E. E., Vessella, R.: The value of serum prostate-specific antigen determinations before and after radical prostatectomy. J. Urol., 141: 873, 1989. 8. Stein, A,, deKernion, J. B., Dorey, F.: Prostate specific antigen related to clinical status 1 to 14 years after radical prostatectomy. Brit. J. Urol., 67: 626, 1991. 9. Hudson, M. A,, Bahnson, R. R. and Catalona, W. J.: Clinical use of prostate specific antigen in patients with prostate cancer. J. Urol., 1 4 2 1011, 1989. 10. Vessella, R. L., Noteboom, J., Lange, P. H.: Evaluation of the
1211
automated Abbott IMx prostate-specific antigen immunoassay. Clin. Chem., 3 8 2044, 1992. 11. Frazier, H. A., Robertson, J . E., Humphrey, P. A. and Paulson, D. F.: Is prostate specific antigen o f clinical importance in evaluating outcome aRer radical prostatectomy. J. Urol., 149: 516, 1993. 12. Montie, J. E.: Follow up after radical prostatectomy or radiation therapy for prostate cancer. UroI. Clin. N. Amer., 21: 673. 1994. 13. Prestigiacomo, A. F., Stamey, T. A.: A comparison of 4 ultrasensitive prostate specific antigen assays for early detection of residual cancer after radical prostatectomy. J. Urol., 152: 1515, 1994. 14. Stamey, T. A., Graves, H. C., Wehner, N., Ferrari, M. and Freiha, F. S.: Early detection of residual prostate cancer after radical prostatectomy by an ultrasensitive assay for prostate specific antigen. J. Urol., 149: 787, 1993. 15. Ferguson, R. A,, Yu, H., Kalyvas, M., Zammit, S. and Diamandis, E. P.: Ultrasensitive detection of prostate-specific antigen by a time-resolved immunofluorometric assay and the Immulite immunochemiluminescent third-generation assay: potential applications in prostate and breast cancers. Clin. Chem., 42: 675, 1996. 16. Mione, R., Barichello, M., Sartorello, P., Leon, A., Barioli, P. and Gion, M.: Third-generation PSA: ultrasensitive or ultraprecise assay? Int. J . Biol. Markers, 1 0 229, 1995. 17. Nee, G. G., Preissner, C. M. and Oesterling, J. E.: Development of a highly sensitive immunochemiluminometric assay for prostate-specific antigen. Urology, 44:76, 1994. 18. Takayama, T. K., Vessella, R. L. and Lange, P. H.: A brief review of ultrasensitive prostate-specific antigen assays for the evaluation of patients after radical prostatectomy. World J. Urol., 11: 192, 1993. 19. Yu, H. and Diamandis, E. P.: Ultrasensitive time-resolved immunofluorometric assay of prostate-specific antigen in serum and preliminary clinical studies. Clin. Chem. 39: 2108, 1993. 20. Graves, H. C., Wehner, N. and Stamey, T. A.: Ultrasensitive radioimmunoassay of prostate-specific antigen. Clin. Chem.. 3 8 735, 1992. 21. Pruthi, R. S., Haese, A,, Huland, E. and Stamey, T. A,: Use of serum concentration techniques to enhance early detection of recurrent prostate cancer after radical prostatectomy. Urology, 4 9 404, 1997. 22. Huland, E., Haese, A. and Huland, H.: Supersensitive detection of prostate-specific antigen in lyophilisation-concentrated serum: a rapid and simple procedure. Int. J. Oncol., 9 593, 1996. 23. van Iersel, M. P., Thomas, C. M., Segers, M. F., Witjes. W. P., Debruyne, F. M. and Oosterhof, G. 0.:The use of 'ultrasensitive' prostate-specific antigen assays in the detection of biochemical recurrence after radical prostatectomy. Brit. J.Urol., 77: 418, 1996. 24. Yu, H., Diamandis, E. P., Wong, P. Y., Nam, R. and Trachtenberg, J.: Detection of prostate cancer relapse with prostate specific antigen monitoring at levels of 0.001 to 0.1 microG./L: J. Urol., 157: 913, 1997. 25. Yu, H., Diamandis, E. P., Prestigiacomo. A. F. and Stamey. T. A,: Ultrasensitive assay of prostate-specific antigen used for early detection of prostate cancer relapse and estimation of tumordoubling time after radical prostatectomy. Clin. Chem. 41: 430, 1995. 26. Witherspoon, L. R. and Lapeyrolerie, T.: Sensitive prostate specific antigen measurements identify men with long diseasefree intervals and differentiate aggressive from indolent cancer recurrences within 2 years after radical prostatectomy. J. Urol., 157: 1322. 1997. 27. Link, P., Freiha, F. S. and Stamey, T. A,: Adjuvant radiation therapy in patients with detectable prostate specific antigen following radical prostatectomy. J. Urol.. 145 532, 1991.
Vol. 161, 1206-1211, April 1999 Printed in U.S.A.
THE JOVRVAL OF UROLOGY Cop-mght 0 1999 by AMERICA\ U H O ~ I C A.SSOCUTIO.U, AL Isc
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN IN THE FOLLOWUP OF 422 PATIENTS AFTER RADICAL PROSTATECTOMY ALEXANDER HAESE, EDITH HULAND, MARKUS GRAEFEN, PETER HAMMERER, JOACHIM NOLDUS AND HARTWIG HULAND From the Department of Urology, Unillersity Clinic Eppendorf, Hamburg, Germany
ABSTRACT
Purpose: We validated our ultrasensitive prostate specific antigen (PSA) assay based on lyophilization and 4-fold concentration of patient sera with the clinical long-term followup and according to histopathological characteristics of 422 patients treated with radical retropubic prostatectomy for prostate cancer. Materials and Methods: Each serum sample was divided into 2 aliquots for standard and 4-fold concentrated (ultrasensitive) detection. Samples were analyzed by the same unmodified DPCImmulite* PSA assay. Biochemical relapse was defined as a n increase of at least 0.10 ng./ml. in native serum (equivalent to 0.025 ng./ml. in concentrated serum). Mean followup was 449 days (range 29 to 2,057 ). Kaplan-Meier analysis of standard and ultrasensitive detection results was done, and findings were correlated with pathological stage, Gleason grade, total cancer volume, Gleason grade 4 cancer volume and margin status. Significance of earlier detection in ultrasensitive versus standard detection w a s calculated with the log rank (Mantel-Cox)test with p c0.05 considered significant. Results: Of 442 patients 88 (20.8%)experienced biochemical recurrence. Of this cohort 28 (31%) demonstrated early failure on the ultrasensitive assay which was later confirmed on the standard assay, 37 (42%)had failure simultaneously on both assays and 23 (26%)had failure on the ultrasensitive but remained disease-free on the standard assay. Average time for ultrasensitive assay detection of recurrence was 288 days (standard 555). Kaplan-Meier analysis revealed significant advantages in earlier detection of recurrence with the ultrasensitive assay, and close correlation with pathological stage, Gleason grade, margin status and Gleason grade 4 cancer volume. Time advantages of ultrasensitive versus standard detection were greater for advanced cancers (pTSa/b or greater, Gleason 3 + 4 or greater) than for small, low grade tumors. All patients who had positive results on the standard assay had a previous (28) or simultaneous (37) positive ultrasensitive result. With standard detection 25% of all relapses were evident within the first year of surgery and with ultrasensitive detection the percentage increased to 85.7%. On both assays 334 patients remained free of biochemical recurrence. Conclusions: Our ultrasensitive PSA assay is useful for early detection of biochemical relapse after radical retropubic prostatectomy. It not only provides the same accuracy as conventional PSA assays but also offers the advantage of detecting recurrence about 300 days earlier. Thus, long-term results of radical retropubic prostatectomy series can be calculated sooner. The clinical impact of this assay will be obvious once curative treatment options are available if applied a t the earliest time of evident tumor recurrence. KEY WORDS:prostate-specific antigen, prostatic neoplasms, prostatectomy About 100,000 men undergo radical prostatectomy each year in the United States for clinically localized prostate cancer.' About 30 to 50% of these patients ultimately have biochemical evidence of failure measured by prostate specific antigen (PSA), an organ specific biochemical marker of recurrent disease.Z-" When first generation assays with a PSA detection limit of 0.2 to 0.6 ng./ml. were used, PSA recurrence usually preceded clinical recurrence by 2 to 5 A newer generation of assays, t h e so-called ultrasensitive assays, lowered the PSA detection limit by a factor of about 5 to 10, permitting earlier Work on assays has focused on increasing this sensitivity by lowering PSA detection limits.'4-") An alternative approach is the manipulation Accepted for publication November 13, 1998. hesented at annual meet,ng of ~ ~Association, ~ San Diego. California, May 30-June 4, 1998. * Diagnostic Product Corp.. Los Angeles. California.
of the serum sample. We recently described an ultrasensitive method to detect PSA at followup after radical prostatectomy based on concentration of serum samples.21.'* The validity of PSA values below the detection limit of a particular assay can be tested against background with t h e same (unmodified) assay. True PSA values below t h e detection limit should concentrate above the detection limit and, thus, can be identified, whereas background noise theoretically should not concentrate. By identifying true positive PSA values below the detection limit we can detect PSA recurrence earlier. We validate the clinical applicability of serum concentration in a large series of patients observed for up to 5 years to determine whether this technique would result in earlier detection of recurrent prostate cancer after radical prostatectomy. For clinical validity we correlated of ultrasensitive l ~ ~ , t h e results ~ ~ ~ and standard detection to histopathological features of prostate specimens.
1206
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN MATERIALS AND METHODS
Patient selection and evaluation. We prospectively analyzed postoperative sera from 422 consecutive patients who underwent pelvic lymphadenectomy and radical retropubic prostatectomy for clinically localized prostatic carcinoma. All lymph nodes were negative for prostate cancer metastasis on pelvic lymph node dissection. Patients who received preoperative hormonal therapy were excluded from study. Patients were seen during regularly scheduled visits at our outpatient department, and followup began at 8 to 10 weeks after surgery. Patients typically were seen after 6, 12, 18 and 24 months, and then at 1-year intervals. A total of 1,046 serum samples were analyzed and the number of followup samples ranged from l(176 patients) to 12 (1). Ofthe 422 patients 246 (58.3%) had at least 2 consecutive measurements during followup. Followup ranged from 29 to 2,057 days or 5.6 years (mean 449, median 327 days). Length of followup was less than 1 year in 224 patients (53.1%),more than 1 and up to 3 years in 154 (36.5%),more than 3 and up to 5 years in 40 (9.5%),and more than 5 years in 4. Principles of lyophilization and concentration. Consecutive serum samples from each patient were drawn at each visit before digital rectal examination and transrectal endosonography. Patient sera were brought to our laboratory and divided into 2 aliquots of 1 ml. each (samples 1 and 2). Both samples were frozen a t -8OC. Sample a was left untreated and sample 2 was lyophilized, which resulted in a 10% dry weight powder of the original sample. A standardized quantity of 150 p1. distilled water was added, which resulted in a 250 pl. 4-fold concentrated serum sample. PSA detection. For PSA analysis we used an Immulite PSA assay, which is a solid phase, 2-site chemiluminescent immunometric assay with a polyclonal-monoclonal antibody combination. Analytical PSA sensitivity is 0.04 ng./ml., whereas clinical sensitivity is 0.10 ng./ml.** The untreated (native) and concentrated (ultrasensitive) sera were analyzed by the conventional assay with standard reagent kits and standard master curves. Native and concentrated samples were measured in triplicate. Study design. We continuously monitored PSA after radical prostatectomy in each patient, and compared the results of native and ultrasensitive detection of each sample. Results of both assays showing no PSA recurrence (defined as an increase from less than 0.10 to at least 0.10 ng./ml.) were noted on a specific day as negative or censored. Ultrasensitive value exceeding our definition of positivity (0.10 ng./ml.) but PSA less than 0.10 ng./ml. for native detection was noted as uncensored and censored for ultrasensitive and native samples, respectively. If on a later measurement both samples exceeded 0.10 ng./ml., the native sample was termed uncensored as well. The time difference between the appearance of an uncensored value on the concentrated sample in comparison to the native sample was the time advantage that could be achieved. Recurrence was correlated with various characteristics, including pathological stage, Gleason grade, margin status, cancer volume and Gleason grade 4 cancer volume. For all patients results for native and concentrated sera and significance tests and analyses of PSA-free survival were compared. Histological characterization of the prostate. The prostate was prepared according t o the Stanford protocol. It was inked over the entire surface, fixed in formalin for at least 24 hours and processed with a 3 mm. step section technique. The Gleason system was used for histological grading and the TNM classification was used for staging. Cancer volume was calculated with a computer assisted volumetric program developed at our department of pathology. Additional histological characterization included margin status, total cancer volume and Gleason grade 4 cancer volume. Statistical analysis. To calculate the significance of time
1207
differences between the appearance of uncensored values in standard versus ultrasensitive detection we performed a Kaplan-Meier analysis of PSA-free survival for native and concentrated samples separately. The 2 curves were superimposed for maximal information gain. Significance was calconsidered culated with the Mantel-Cox test with p ~ 0 . 0 5 significant. RESULTS
Of 442 patients 88 (20.86) experienced biochemical recurrence. Of this cohort 28 (31%)demonstrated early failure on the concentrated assay which was later confirmed on the standard assay, 37 (42%)had failure simultaneously on both assays and 23 (26%) had failure on the concentrated but remained disease-free on the standard assay. This study involved a dynamic process and so the confirmation of biochemical failure, defined as an increase in PSA to at least 0.10 ng./ml. in native samples, particularly in the last group of patients, must be awaited. Therefore, evaluation of our ultrasensitive PSA analysis was based on patients who had a positive result first on concentrated and later on native samples. Of 88 patients 28 fulfilled these criteria (table 1).Mean time to evidence of relapse was 555.2 days (range 129 to 1,389), for these patients, whereas the ultrasensitive assay detected PSA relapse a mean of 288 days (range 47 to 1,130) after radical prostatectomy. Thus, a mean lead time of 267 days (range 66 to 650) was possible, which has a great potential benefit for patients. PSA relapse was detected on native samples within the first year after radical prostatectomy in 7 of 28 patients (25%).The ultrasensitive assay detected PSA relapse within the first year in 24 of the 28 patients (85.7%).On both assays 334 patients remained free of biochemical recurrence. The relative frequency of failure compared with pathological stage, Gleason score, total cancer volume, Gleason grade 4 cancer volume and margin status is shown in table 2 for standard and ultrasensitive detection, and for the 23 patients who had positive results with ultrasensitive but negative results with standard detection.
TABLE1 Summary of 28 patients uvth positiLv resuits rnrtrallv on ultrasensitive and later on standard PSA assajb First Pos Result tdaj)
Pt No
~
Native
1 2 3 4 5 6 7
557 368 1,010 408 1.192 378 556 1,395 373 3 75 572 196* 476 581 728 656 720
Concentrated
- -
Dtflerence
249 193* 1 75 573 437 209* 199 360 832 193' 185 365' 191 59s 8 800 176 9 197* 160' 215 10 215 357* 11 126 12 70* 211 13 265* 407 174' 14 650 15 78' 327 16 323* 488 17 232' 367 303 18 64* 217* 126 91' 19 96 206302* 20 259 1,389 1.130 21 213 287* 72' 22 140 202* 34'" 23 66 129* 63. 24 282 595 313' 25 210 535 325* 26 364 242* 606 27 195 47* 28 242' __ Mean (range) 555 (129-1.389) 288 (47-1.130) 267 (66-650) * Biochemical failure 1 year or less aRer surgery on standard assay in 7 of 28 patients (254%)and on ultrasensitive assay in 24 of 28 patients (85.7%). 308'
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN
1208
TABLE2. Comparison of histopathological characteristics and PSA relapse on ultrasensitiue and natiue assays No. Ultrasensitive Assay Pos. Only ( 7 6 ) No. Both Assays Pos. (B) Total No. Pts. Pathological stage: pT2a-c
2 (1.11 25 (20.6) 33 (47.81 5 (50) 65
3 (1.6) 12 (9.9) 8 111.59)
136 163 62 13
1 (0.71 23t14.1) 31 (50.0) 10 (76.9) 65
0 10
(6.1) 11 (17.71 -2 (15.41 23
69 297
29 (42.2) 33 (11.0) 62
4 16 20
0 3 (5.4) 13 f 15.9) 12 (20.7) 24 (44.41 52
0
185 121 69 10
pT3ah
pT3c PT4 Totals Gleason score: 3 + 3 or Less
314 4+3 4 + 4 or Greater Totals
Margin status: Pos. Neg. Totals Tumor vol. (cc): Less than 0.5
22 56 82 58 54
0.5-2 2-5 5-10 Greater than 10 Totals
- -
23
(5.8) (5.41
0 5 (6.1) 10 (17.21 2 (3.7) 17
Gleason grade 4 tumor vol. (ccl:
0
117 128 73 30 40
Less than 0.5 0.5-2
2-5 Greater than 5
0
4 (3.1) 13 (17.8) 12 (40.01 23 (57.51 52
- _ Totals _ _ _ ~ Kaplan-Meier curves for PSA-free survival are shown in figure 1 for all patients (p = 0.03). These results were stratified according to histopathological features of specimens. The respective Kaplan-Meier curves in figure 2 show significance according to the log rank (Mantel-Cox) test. Comparison of standard and ultrasensitive detection revealed significant advantages in favor of the latter for pathological stage margin status (p = 0.03) (p = 0.016), Gleason grade (0.007), and Gleason grade 4 cancer volume ( p = 0.027). Since there was no recurrence in patients with less than 0.5 cc total cancer volume, a p value could not be calculated for this parameter. To determine the time benefit of ultrasensitive detection of PSA we created a Kaplan-Meier curve that included all patients who had positive results for native and concentrated sera. Patients with positive results on ultrasensitive but negative results on standard detection were excluded from analysis. Figure 3 shows the time advantage expressed as the area between curves (p = 0.04). DISCUSSION
The first generation PSA assays used clinically had a detection limit of 0.2 to 0.6 ng./ml. However, for followup of
1 ,
-
..
.e Cum SuMval ( a d ) Cum Survival (sup)
u
.2
'
-Chi-square 4,655-
CF-
P-Value.
1 ,
,0310
.-
0-., 0
500
1000
1500
2000
Time
FIG. 1. Kaplan-Meier curve shows PSA-free survival in standard (std)and ultrasensitive (sup)detection for all patients. Cum., Cumu-
lative.
0 1 (0.78) 2 (2.71 7 (23.3) 10 (25) 20
radical prostatectomy patients a lower detection limit of 0.2 to 0.6 ng./ml. proved insufficient for the minute changes in serum PSA that indicated recurrence of prostate cancer. Therefore, various efforts have been made to improve the sensitivity of PSA assays to reduce the time between radical prostatectomy and evidence of biochemical relapse. In 1992 Graves et a1 reported a successful method which increased the analytical sensitivity of the standard assay from a working range of 0.3 to 50 to 0.1 to 1.2 ng./ml. with a modification of a radioimmunoassay.20 PSA values in 2 patients with recurrent prostatic cancer increased above the 0.1 ng./ml. detection limit of the modified assay 175 and 581 days before reaching the 0.3 ng./ml. detection limit of the standard assay, which demonstrates a significant time advantage of potential benefit to patients with respect to beginning adjuvant therapy earlier. In the screening of 187 post-radical prostatectomy patients for whom the standard assay revealed no evidence of cancer the ultrasensitive assay described by Graves et a120 was applied by Stamey et all4 and detected evidence of residual cancer (PSA at least 0.1 ng./ml.) in 21 patients (11.2%).In the same study simultaneous analysis with both assays was performed on sera of 22 patients with proved recurrence. The ultrasensitive assay (sensitive to PSA at at least 0.1 ng./ml.) detected PSA recurrence much earlier in all 22 patients than the standard assay (sensitive to PSA at least 0.3 ng./ml.) by a median and mean of 202 and 310 days, respectively. The technical backbone of the ultrasensitive assay was a longterm incubation of samples and the creation of a standard curve which was adjusted to the extremely low PSA values in post-radical prostatectomy sera. This modification precluded the use of the assay in the followup of patients with low serum PSA concentrations. Yu and Diamandis reported an ultrasensitive method for measuring PSA in serum, including a monoclonal-polyclonal antibody combination and alkaline phosphatase labeled ~treptavidin.'~ The main characteristic was sensitivity as it could detect PSA as low as 0.002 nglml. Based on a monoclonalmonoclonal combination and 1-step incubation as a further improvement Ferguson et a1 demonstrated a lower limit of
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN
1209
Cum Survival (llsld)
cum survival (1lMd) Cum Survival (llarp) Cum Survival (ZIsldJ
Cum. Survival (Ilsup) Cum. Survival (ZIed)
Cum Survival (2/supJ
Cum. Survival (Zsup) Cum. Survival (Ysld)
g
Cum. Survival (Ysup)
0
1000
500
1500
,4
o
cum Sumval (3mj
-1
Cum Survival (Yaup)
i
Cum Survival (41816)
Cum survnval (4lsld)
Cum. Survival (Usup)
cum SuMval (4lS"P)
0
2000
500
Time
1000
1500
2000
TlmR
D Cum. Survival (yee/¶d) Cum. Sumval (yedsup)
g 4
1iirr' Chi-Sqare
0
Cum Survival (noled)
L
-_
4,368
Cum. Survival (ndsup)
F P-Value
,0366
!
1
_?__i
500
0
I000
1500
2000
Time
E
-
'
~-
14
L-
Cum. Survival (zerolsld) Cum. survival Izom/sup) Cum Survival (
1
1 t
i o
500
iooo
1500
Cum survival (0.5-2/sup) Cum. Survival (,2-5/sId) Cum Survival (>Z-SIsup) Cum. Survival (>wnd) cum. survivai (>%sup)
zoo0
Time
FIG. 2. Kaplan-Meier curves. A, PSA-free survival in standard (std) and ultrasensitive (sup)detection for pathological stages 2a to c ( I ) , 3ah (2),3c ( 3 )and 4 ( 4 ) . B,PSA-free survival in standard and ultrasensitive detection for Gleason scores 3 + 3 or less ( I ) , 3 + 4 ( Z ) , 4 + 3 (3) and 4 + 4 or greater ( 4 ) . C , PSA-free survival in standard and ultrasensitive detection for positive margin (yes) and negative margin. D , PSA-free survival in native and ultrasensitive detection for total tumor volume in 5 groups. Because there was no recurrence in group with less than 5 cc tumor volume log rank (Mantel-Cox)test could be calculated. E , PSA-free survival in standard and ultrasensitive detection for Gleason grade 4 cancer volume in 5 volume groups. Cum., Cumulative.
'
Cum. Survival (std) Cum. Survival (sup)
t 0
400
800
1200
1600
Time
FIG. 3. Kaplan-Meier curve shows PSA-free survival in standard (std) and ultrasensitive (sup)detection for 65 patients with ultrasen-
sitive and standard evidence of biochemical failure. Area between curves represents time benefit achieved by lyophilization concentration. C u m . , Cumulative.
detection of better than 0.001 ngfml.15 Comparison with a commercially available third generation assay revealed substantial PSA changes at concentrations undetectable with conventional assays. A retrospective study with the ultrasensitive method of Ferguson et a l l 6 was performed by Yu et al. in 148
patients with a postoperative PSA of less than 0.1 ngfml. according to a conventional PSA assay.**Increases in postoperative serum PSA at levels of 0.001 to 0.1 ngfml. after radical prostatectomy were associated with clinicopathologicalfeatures of poor prognosis, and they concluded that monitoring patients with a highly sensitive PSA assay (detection limit 0.001 ngfml.) was a simple and effective method of detecting clinically important biochemical relapse soon after radical prostatetomy. This was one of the first studies to correlate biochemical failure on extremely low concentrations of serum PSA with histopathological features of prostate cancer. Positive margins (p <0.01), increasing tumor volumes ( p <0.01), higher preoperative PSA (p = 0.03) and extracapsular penetration (p = 0.02) were significantly associated with recurrence, and recurrence could be detected by measuring PSA at concentrations not detected by standard assays. Our approach to increase the sensitivity of PSA assays is different. Rather than modifying the assay specifically, for example by increasing the incubation time or developing a master curve adjusted to low levels of PSA described by Graves et a1,20 improving antibodies or introducing monoclonal-monoclonal combinations and so forth, we focused on the serum sample. This approach has the advantage of leaving well established assay designs untouched, permits
1210
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN
the application of our technique in each laboratory without large investments, and permits post-radical prostatectomy samples to be analyzed with the normal standard assay and even in the same test run with routine samples of preoperative sera. The serum concentration techniques were first described by us.22 In another study Pruthi et al retrospectively analyzed sera by 2 concentration modes of lyophilization and ultrafiltration.21 Sera from women, cystoprostaktomy patients, those cured by radical prostatectomy, those with benign prostatic hyperplasia and those whose prostate cancer was treated with radiation or hormonal therapy were used as controls. The primary study group consisted of 31 patients who had recurrent disease after radical prostatectomy and whose initial postoperative PSA decreased to undetectable levels (less than 0.07 ngJml.) and later was positive (greater than 0.07 ngJml.) according to the TOSOH AIA* 600 immunoassay using the ultrasensitive mode. Sera concentrated by lyophilizationand filtration demonstrated PSA recurrence significantly earlier than unconcentrated sera in 29 (94%)and in 28 (90%)of the 31 patients, respectively. Mean time advantage was 362 days for 29 patients and 383 for 28, which correlated well with our results of 308 to 366 days.2" Sera from women, cystoprostatectomy patients and those cured by radical prostatedomy failed to concentrate, yielding a test specificity of 100%as in our study.22 Pruthi et al concluded that serum concentration, irrespective of the method used, is a specific and sensitive technique which provides a significant lead time of about 12 months to detect cancer recurrence after radical prostatectomy. More important, the combination of serum concentration with a more advanced assay (the AIA) gave an additional lead time of 9 months at a 0.07 ngJml. residual cancer detection limit, which yielded a total benefit of about 2 years compared with first generation assays which had a 0.2 ngJml. residual cancer detection limit (Hybritech Tandem-R? assay). In 1995 Yu et al reported on ultrasensitive PSA detection (0.02 ngJml. lower limit) to monitor PSA changes in 10 patients in whom prostatectomy failed.25Prostatic cancer recurrence could be diagnosed a mean of 420 or 883 days earlier with the ultrasensitive assay than with assays that had a 0.1 or 0.3 ngJml. detection limit, respectively. They suggested that more sensitive PSA assays might be able to reduce the relapsefree periods further in about 50% of the prostatic cancer patients who ultimately have relapse. Another study comparing a first generation assay (Tandem-Et Singlepint PSA assay) with improved successors (IMxS and Tandem-E Multipoint PSA assays) was done by van Iersel et al in 1996.23 Compared with the lead time obtained with the Tandem-E Singlepoint assay, 12 of 137 patients with biochemical recurrence had a median gain in lead time of 327 days (range 60 to 627) with the IMx and 369 days (range 60 to 639) with the Tandem-E Multipoint assays. It is by rational interpretation and continuation of these studies that further improvement of future assay generations will decrease the time needed to detect recurrence of prostate cancer. However, because our method is independent of the assay itself, that need not restrict the use of serum concentration. When we study the time differences in detection between standard and ultrasensitive methods, a striking feature is that not all patients seem to benefit from ultrasensitive detection. A relatively large number of patients (37 of 88) had positive results with standard and ultrasensitive detection simultaneously despite previous negative results with both assays. In these patients there might have been a time at which the ultrasensitive assay could have detected PSA and the standard assay could not but this time was missed because of the limitations of blood sampling intervals during followup. We are aware that some patients had a time advantage with ultrasensitive detection that might well have been smaller if the sampling intervals had been shorter.
* TOSOH Medics, Foster City, California.
t H britech, Inc., San Diego. California. $ Albott Laboratories, Abbott Park, Illinois.
The relatively short mean followup of our study in our opinion does not affect our results because individual length of followup is essential. If a patient has recurrent disease relatively soon after radical prostatectomy, that is within 1 year after surgery as in 24 of 28 patients in our study, the need for longer followup is less important. On the other hand, we have patients with followup greater than 3 and 5 years, and most are free of disease. We did not evaluate the outcome of radical prostatectomy but the ability of 2 assay types to detect PSA indicative of disease recurrence. Thus, the comparison of the time differences at which recurrence is detected by the assays is important, rather than the absolute postoperative followup length before recurrence. The correlation of pathological features of the prostate specimen and results of standard and ultrasensitive detection demonstrate that, irrespective of the feature examined (for example pathological stage), the time to detect a recurrence was significantly earlier for ultrasensitive versus standard detection (fig. 2). As described by others advanced pathological stage, predominance or even presence of Gleason grades 4 and 5, and so forth in the prostate are typically associated with PSA levels that can be detected earlier using assays with increased s e n s i t i ~ i t y . ~ ~ . ~ ~ In our cohort of patients who ultimately had radical prostatectomy failure as proved by an increase in PSA from undetectable to detectable levels initially on ultrasensitive and subsequently on standard detection (281, the PSA levels were continuously increasing to levels higher than 0.10 ng./ ml. Thus, we assume it is correct to classify those cases as radical prostatectomy failure. Another 23 patients in the study group had biochemical failure on ultrasensitive detection and the increase in PSA to detectable levels on standard measurements is awaited. However, we expect at least a large number of these patients to have failure at some time, in particular those with unfavorable pathological conditions. As we have shown in table 2 the patients who have already had failure compared to those with no detectable PSA on standard measurement have a comparable distribution of pathological features. As noted previously we observed a dynamic process, and so it is not surprising that 23 patients have no proved recurrence yet. Due t o shorter followup the reconfirmation of failure has not occurred to date. The potential extraprostatic source of PSA (periurethral glands) seemed to be of no relevance in patients who had failure according to PSA detection. If PSA from those sources had mimicked PSA failure, it should have been detectable on the first postoperative serum sample. Also, in this case the level of PSA should be relatively stable at low concentrations rather than continuously increasing to levels indicative of recurrence which occurred in our study. The clinical relevance of ultrasensitive detection of PSA can be explained by its impact on the accurate assessment of the outcome after radical prostatectomy and the treatment of patients. Patients who are at risk for failure due to unfavorable pathological conditions should be monitored more closely, particularly if ultrasensitive detection is suggestive of failure. On the other hand, patients with a pathological condition that is less suggestive of recurrence and no sign of PSA recurrence on ultrasensitive detection can be followed less extensively. The established treatment modalities for recurrent prostate cancer, that is hormonal and radiation therapy, offer no cure. Moreover, in other malignancies therapies for recurrent disease offer advantages in terms of cure when instituted at the earliest time. The efficacy of early treatment of recurrent prostate cancer has not been investigated with long-term followup to our knowledge.24,~Ultrasensitive detection of PSA provides an improved tool to identify patients who might benefit from those therapies. In particular patients who are relatively young at the time of proved recurrence should be monitored intensively.
ULTRASENSITIVE DETECTION OF PROSTATE SPECIFIC ANTIGEN CONCLUSIONS
Ultrasensitive PSA detection i s useful for early detection of biochemical relapse after radical retropubic prostatectomy. It not only provides the s a m e accuracy as conventional PSA a s s a y s b u t also offers the a d v a n t a g e of yielding a positive r e s u l t a b o u t 300 d a y s earlier. Most biochemical failures can be detected within the first y e a r a f t e r radical prostatectomy. Sensitivity and specificity are equal t o those of standard PSA assays. Correlation of ultrasensitive assay findings with pathological f e a t u r e s i s equal t o that of the standard assay, which i s a sufficient rationale to apply t h i s modification at o u r laboratory. Independence of serum concentration from the a s s a y t y p e p e r m i t s use w i t h present and f u t u r e generation a s s a y s w i t h o u t questioning these improved assays. Thus, t h i s m e t h o d is a completion rather t h a n a competitor of new generation assays or a s s a y types developed b y others.19.2" An optimized PSA a s s a y m a k e s e v e n m o r e sense in combination w i t h an optimized s e r u m s a m p l e t o achieve m a x i m a l information w i t h respect t o detecting the minute changes in PSA concentrations that indicate PSA recurrence a f t e r radical prostatectomy. Long-term results of radical retropubic prostatectomy series c a n be calculated earlier. The clinical i m p a c t of t h i s a s s a y will be obvious a f t e r curative t r e a t m e n t options are available if it is used at the earliest t i m e of evident t u m o r recurrence.
REFERENCES
1. Olsson, C. A. and Goluboff, E. T.: Detection and treatment of prostate cancer: perspective of the urologist. J . Urol., 1 5 2 1695, 1994. 2. Partin, A. W., Pound, C. R., Clemens, J. Q., Epstein, J. I. and Walsh, P. C.: Serum PSA after anatomic radical prostatectomy. The Johns Hopkins' experience after 10 years. Urol. Clin. N. Amer., 2 0 713, 1993. 3. Trapsasso, J. G., deKernion, J. B., Smith, R. B. and Dorey, F.: The incidence and significance of detectable levels of serum prostate specific antigen after radical prostatectomy. J. Urol., 152: 1821, 1994. 4. Stamey, T. A., Yang, N., Hay, A. R., McNeal, J . E., Freiha, F. S. and Redwine, E. A.: Prostate-specific antigen as a serum marker for adenocarcinoma of the prostate. New Engl. J. Med., 317: 909, 1987. 5. Oesterling, J. E., Chan, D. W., Epstein, J. I., Kimball, A. W., Bmzek, D. J., Rock, R. C., Brendler, C . B. and Walsh, P. C.: Prostate-specific antigen in preoperative and postoperative evaluation of localized prostatic carcinoma treated with radial prostatectomy. J. Urol., 1 3 9 766, 1988. 6. Stamey, T. A,, Kabalin, J. N., McNeal, J. E., Johnstone, I . M., Freiha, F. S., Redwine, E. A,, Yang, N.: Prostate specific antigen in the diagnosis and treatment of adeno-carcinoma of the prostate: 11. Radical prostatectomy treated patients. J. Urol., 141: 1076, 1989. 7. Lange, P. H., Ercole, C. J., Lightner, D. J., Fraley, E. E., Vessella, R.: The value of serum prostate-specific antigen determinations before and after radical prostatectomy. J. Urol., 141: 873, 1989. 8. Stein, A,, deKernion, J. B., Dorey, F.: Prostate specific antigen related to clinical status 1 to 14 years after radical prostatectomy. Brit. J. Urol., 67: 626, 1991. 9. Hudson, M. A,, Bahnson, R. R. and Catalona, W. J.: Clinical use of prostate specific antigen in patients with prostate cancer. J. Urol., 1 4 2 1011, 1989. 10. Vessella, R. L., Noteboom, J., Lange, P. H.: Evaluation of the
1211
automated Abbott IMx prostate-specific antigen immunoassay. Clin. Chem., 3 8 2044, 1992. 11. Frazier, H. A., Robertson, J . E., Humphrey, P. A. and Paulson, D. F.: Is prostate specific antigen o f clinical importance in evaluating outcome aRer radical prostatectomy. J. Urol., 149: 516, 1993. 12. Montie, J. E.: Follow up after radical prostatectomy or radiation therapy for prostate cancer. UroI. Clin. N. Amer., 21: 673. 1994. 13. Prestigiacomo, A. F., Stamey, T. A.: A comparison of 4 ultrasensitive prostate specific antigen assays for early detection of residual cancer after radical prostatectomy. J. Urol., 152: 1515, 1994. 14. Stamey, T. A., Graves, H. C., Wehner, N., Ferrari, M. and Freiha, F. S.: Early detection of residual prostate cancer after radical prostatectomy by an ultrasensitive assay for prostate specific antigen. J. Urol., 149: 787, 1993. 15. Ferguson, R. A,, Yu, H., Kalyvas, M., Zammit, S. and Diamandis, E. P.: Ultrasensitive detection of prostate-specific antigen by a time-resolved immunofluorometric assay and the Immulite immunochemiluminescent third-generation assay: potential applications in prostate and breast cancers. Clin. Chem., 42: 675, 1996. 16. Mione, R., Barichello, M., Sartorello, P., Leon, A., Barioli, P. and Gion, M.: Third-generation PSA: ultrasensitive or ultraprecise assay? Int. J . Biol. Markers, 1 0 229, 1995. 17. Nee, G. G., Preissner, C. M. and Oesterling, J. E.: Development of a highly sensitive immunochemiluminometric assay for prostate-specific antigen. Urology, 44:76, 1994. 18. Takayama, T. K., Vessella, R. L. and Lange, P. H.: A brief review of ultrasensitive prostate-specific antigen assays for the evaluation of patients after radical prostatectomy. World J. Urol., 11: 192, 1993. 19. Yu, H. and Diamandis, E. P.: Ultrasensitive time-resolved immunofluorometric assay of prostate-specific antigen in serum and preliminary clinical studies. Clin. Chem. 39: 2108, 1993. 20. Graves, H. C., Wehner, N. and Stamey, T. A.: Ultrasensitive radioimmunoassay of prostate-specific antigen. Clin. Chem.. 3 8 735, 1992. 21. Pruthi, R. S., Haese, A,, Huland, E. and Stamey, T. A,: Use of serum concentration techniques to enhance early detection of recurrent prostate cancer after radical prostatectomy. Urology, 4 9 404, 1997. 22. Huland, E., Haese, A. and Huland, H.: Supersensitive detection of prostate-specific antigen in lyophilisation-concentrated serum: a rapid and simple procedure. Int. J. Oncol., 9 593, 1996. 23. van Iersel, M. P., Thomas, C. M., Segers, M. F., Witjes. W. P., Debruyne, F. M. and Oosterhof, G. 0.:The use of 'ultrasensitive' prostate-specific antigen assays in the detection of biochemical recurrence after radical prostatectomy. Brit. J.Urol., 77: 418, 1996. 24. Yu, H., Diamandis, E. P., Wong, P. Y., Nam, R. and Trachtenberg, J.: Detection of prostate cancer relapse with prostate specific antigen monitoring at levels of 0.001 to 0.1 microG./L: J. Urol., 157: 913, 1997. 25. Yu, H., Diamandis, E. P., Prestigiacomo. A. F. and Stamey. T. A,: Ultrasensitive assay of prostate-specific antigen used for early detection of prostate cancer relapse and estimation of tumordoubling time after radical prostatectomy. Clin. Chem. 41: 430, 1995. 26. Witherspoon, L. R. and Lapeyrolerie, T.: Sensitive prostate specific antigen measurements identify men with long diseasefree intervals and differentiate aggressive from indolent cancer recurrences within 2 years after radical prostatectomy. J. Urol., 157: 1322. 1997. 27. Link, P., Freiha, F. S. and Stamey, T. A,: Adjuvant radiation therapy in patients with detectable prostate specific antigen following radical prostatectomy. J. Urol.. 145 532, 1991.