Introduction

Prostate cancer is the second leading cause of
cancer-related deaths in US men. It is estimated that 184,500 new
cases of prostate cancer will be diagnosed in the United States in
1998, and over 39,200 deaths will result from this cancer.[1]
Recently, the National Cancer Institute has determined that the death
rate from prostate cancer has decreased slightly; this may reflect
favorable results achieved through early detection.[2]

Unfortunately, the majority of patients with prostate cancer have
extraprostatic spread at the time of diagnosis. Although nearly 60%
of newly diagnosed cases of prostate cancer are predicted to be
organ-confined based on digital rectal examination (DRE), serum
prostatic specific antigen (PSA) levels, and histologic grading
(Gleason score on biopsy), between 26% and 68% of these patients are
ultimately found to have organ-confined disease on final pathologic
analysis.[3] Thus, more than one-third of men with clinically
localized prostate cancer have extraprostatic disease at the time of
surgery and are not curable by surgery alone.[4]

Currently, preoperative staging of clinically localized prostate
cancer is based on DRE findings, serum PSA level, and the results of
prostate needle biopsy and radiologic explorations. Digital rectal
examination frequently understages the extent of tumor. Imaging
modalities, such as endorectal coil magnetic resonance imaging (MRI)
and ultrasound, add little to the accuracy of preoperative staging.[5]

The preoperative serum PSA value may be difficult to interpret, due
to the volume of benign prostatic hyperplasia and the degree of tumor
differentiation.[4] Therefore, serum PSA levels are not sufficiently
reliable to predict final pathologic stage on an individual basis in
patients with localized prostate cancer.

Preoperative Gleason score determined from biopsy specimens strongly
correlates with final pathologic stage in patients at either extreme
of the scoring system (ie, those with Gleason scores from 2 to 4 or
from 8 to 10).[6] Unfortunately, prediction of pathologic stage is
not as reliable for the more than 75% of men whose Gleason score
falls between 5 and 7. Thus, improvements in existing preoperative
staging techniques are clearly needed.

Recent advances in molecular biology have allowed for the detection
of prostate cells in the peripheral blood. Through the use of reverse
transcription followed by the polymerase chain reaction (RT-PCR) with
primers specific for PSA messenger RNA (mRNA) or prostate-specific
membrane antigen (PSMA) mRNA, circulating prostate cells can be
identified.[7-36] This technique has been reported to detect cells in
over 80% of patients with metastatic disease and in 40% of those with
clinically localized prostate cancer.

There is a considerable debate over the clinical use of RT-PCR as a
preoperative indicator of extraprostatic disease. This article will
review the literature on the ability of RT-PCR to predict final
pathologic stage and PSA recurrence after radical prostatectomy.

RT-PCR as a Predictor of Pathologic Stage

In 1992, Moreno et al reported the detection of circulating cells by
RT-PCR for PSA in patients with metastatic prostate cancer.[33] In
1994, Katz et al introduced the concept of molecular staging using
RT-PCR technology.[7] They found that the test could predict surgical
failure (ie, capsular penetration, surgical margins, and seminal
vesicle involvement) with high specificity and sensitivity.

Since the publication of the study by Katz et al,[7] several teams
have developed RT-PCR procedures to detect circulating cells and have
evaluated these procedures in their patients.[13-36] To date, only
one other team found a correlation between RT-PCR–positive
results and pathologic stage.[13]

Columbia Experience

At Columbia-Presbyterian Medical Center, ongoing investigations have
studied the performance of RT-PCR in the staging of prostate cancer
patients prior to radical prostatectomy.[7-12] The test’s
sensitivity, specificity, positive predictive value, negative
predictive value, accuracy, and odds ratio for various surgical
outcomes are summarized in Table 1.
Patients with a positive RT-PCR assay had a 67% chance of having
extraprostatic extension, a 49% likelihood of surgical failure, and a
60% likelihood of capsular penetration.

Recently, Nejat et al[12] described the stratification of RT-PCR PSA
results according to preoperative serum PSA levels (£
10 ng/mL vs > 10 ng/mL). As shown in Table
2, this stratification can improve staging: 91% of patients with
RT-PCR that was positive for PSA and a preoperative serum PSA level
> 10 ng/mL had pT3 disease, as compared with 36% of patients with
a negative RT-PCR for PSA (chi-square test = 17.2; P = .001). If PSA
stratification was not used, 60% of patients with RT-PCR positive for
PSA had pT3 disease vs 21% of those with RT-PCR negative for PSA.

Literature Review

In our literature review, we divided clinical studies according to
the type of RT-PCR protocol used, as well as the molecular target
(PSA, PSMA, or hK2) (Table 3). The
initial clinical report of Moreno et al describing the detection of
circulating prostate cells involved patients with D0 to D3
disease.[33] The RT-PCR assay was positive in one-third of the cases,
indicating the presence of circulating prostatic cells.

Additional studies examining patients with known metastatic prostate
cancer indicated that RT-PCR detected circulating cells in the
peripheral blood or bone marrow in 31% to 100% of patients (Table
3). The majority of studies consistently failed to detect
PSA-expressing cells in control populations, as shown in the negative
control column in Table 3.

Reverse transcriptase–polymerase chain reaction analyses of
venous blood from patients with clinically localized prostate cancer
(T1-T2) identify hematogenous PSA-expressing cells in 0% to 81% of
patients prior to radical prostatectomy. As shown in Table
3, most groups have reported a gradual increase in
RT-PCR–positivity (regardless of RT-PCR target or sample source)
related to increasing tumor stage (ie, clinically localized vs
metastatic prostate cancer).

In the literature, 1,158 RT-PCR assays for PSA in peripheral blood
have been carried out, 757 in patients with pT1 to pT2 disease and
401 in patients with pT3+ disease. Overall, 23% (174/757) of patients
with organ-confined disease and 38% (151/401) of those with
extraprostatic disease were RT-PCR–positive. All but two groups
reported no advantage to using RT-PCR as a clinical modality at this time.

Recently, Grasso et al described their experience using a nested
RT-PCR assay for PSA and PSMA.[13] They found a statistical
difference in RT-PCR–positivity between patients with pT2
disease and those with pT3 disease (37.5% and 81.5% respectively; P =
.001). Their PSA/PSMA nested RT-PCR had an odds ratio of 7.3 (95%
confidence interval [CI], 2.3 to 23.4) for predicting tumor
extraprostatic extension, consistent with the Columbia University
data (odds ratio: 5.5; 95% CI, 3.2 to 9.3).

Possible Reasons for Differences Among Institutions

There is much debate in the field over the validity and clinical
significance of the detection of circulating prostate cells using PCR
technology. The great disparity in the PCR techniques performed at
different laboratories could account for the varying results reported
in different studies. After 1995, seven academic institutions
established a PCR consortium to accrue at least 300 patients
undergoing radical prostatectomy and assess the role of RT-PCR in
staging and prediction of recurrence.[29] This study confirmed
specimen-related processing, technical, and clinical assay
performance variability issues, indicating that equivalent technical
performance does not equate with clinical performance.

Besides differences in assays, there are differences regarding
selection criteria for patients (such as patient age, preoperative
serum PSA, and a lack of standardized conditions for collection of
samples), technical specifications for running the assay (such as
varying mRNA harvesting conditions, unique primers, and distinct PCR
conditions) and pathologic proceedings. Differences in these
parameters could prevent comparisons of results from different
studies from being made (see Table 4).