The Use of Real-Time Reverse Transcription-PCR for Prostate-Specific Antigen mRNA to Discriminate between Blood Samples from Healthy Volunteers and from Patients with Metastatic Prostate Cancer

Since our original report using the detection of tissue-specific gene expression by reverse transcription-PCR (RT-PCR) to indicate the presence of circulating tumor cells in blood (1), there have been many studies describing the application of RT-PCR in a number of different tumor types. These initial studies used tyrosinase as a marker of melanoma, but many studies have used prostate-specific antigen to detect prostate cancer. Prostate cancer is the most common cancer to affect men with an annual incidence of 72.7 cases per 100,000 in the United Kingdom (2) and 160.4 per 100,000 in the United States (3). Most patients are diagnosed with clinically early stage disease and are presented with a number of treatment options that include surgery, radiotherapy, hormonal therapy, and surveillance. There is a lack of a strong evidence base to support one therapeutic option over another, and decisions regarding treatment are based on stage, age, histologic grade, baseline prostate-specific antigen level, comorbidity, and informed patient choice (4). When considering surgery as a therapeutic option, it is important to identify patients with clinically locally advanced disease (T3 and T4 lesions or nodal involvement) who are not usually amenable to complete excision and who have a higher rate of subsequent relapse. Hence, accurate staging to identify such cases before definitive treatment is important but is often not precise. In this context conventional nonquantitative RT-PCR for the detection of the tissue-specific gene prostate-specific antigen has been extensively studied in blood samples from patients with clinically localized prostate cancer who are about to undergo radical prostatectomy. It was envisaged that such evaluation of peripheral blood before definitive surgery might more accurately predict those patients with pathological higher stage localized disease (T3 and T4 lesions) and at high risk of metastases who are less likely to benefit from surgery. Many studies have reported a positive preoperative RT-PCR result for prostate-specific antigen mRNA to predict for capsular penetration, seminal vesicle involvement, positive surgical margins, and subsequent biochemical relapse (5, 6, 7, 8), but other studies have failed to confirm this (9, 10, 11, 12, 13, 14, 15). The clinical utility of conventional nonquantitative RT-PCR in this setting remains undefined. Importantly, although most of these studies using prostate-specific antigen as a target for RT-PCR have observed absolute specificity, some authors have detected prostate-specific antigen transcripts in blood samples from healthy volunteers (9, 10, 16, 17, 18, 19). We have made similar observations using conventional non-nested RT-PCR for prostate-specific antigen mRNA.³ The detection of prostate-specific antigen mRNA in blood samples from healthy volunteers is most likely due to the activity of tissue-specific promoters called illegitimate transcription (20). Hence, the specificity of RT-PCR for prostate-specific antigen mRNA to detect circulating tumor cells is not absolute. This may be a contributing factor to the conflicting results reported to date using RT-PCR for prostate-specific antigen mRNA to detect circulating tumor cells and highlights the limitation of conventional nonquantitative RT-PCR in discriminating between healthy individuals and patients with disease.

In the present study quantitative investigation of prostate-specific antigen mRNA expression in blood samples from healthy volunteers has been made and compared with levels in blood samples from patients with localized and metastatic prostate cancer. Using such an approach might improve specificity by defining the level of prostate-specific antigen mRNA expression in blood samples from both healthy volunteers and patients with prostate cancer to identify an appropriate cutoff to distinguish these two clinically distinct groups.

Blood samples (2 mL) for RT-PCR were taken from a total of 40 patients with metastatic prostate cancer. This included patients with newly diagnosed untreated metastatic prostate cancer (n = 28) and a heterogeneous group of patients with metastatic prostate cancer (n = 12). The heterogeneous group included patients receiving hormonal treatment intermittently who at the time of blood sampling were off treatment (n = 6), patients with androgen-independent disease (n = 5), and 1 patient responding to hormonal therapy. In a subset of patients with newly diagnosed prostate cancer (n = 14), follow-up blood samples were taken between 3 and 18 months after initiation of hormonal treatment. In addition, blood samples (2 mL) were collected from patients with localized prostate cancer (n = 27); some of these were collected before radical prostatectomy (n = 19). Blood samples (n = 21) were also taken from healthy volunteers (16 males and 5 females), all of who were under 40 years of age.

All of the blood samples (2 mL) were collected into EDTA-containing blood collection tubes (BD Vacutainer) and were immediately added to 8 mL of Ultraspec RNA isolation solution (Biogenesis Ltd., Poole, England) and frozen at −80°C until RNA was extracted. In addition, blood samples (3.5 mL) were also collected into a BD SST tube (containing silica clot activator polymer gel, BD Vacutainer) for prostate-specific antigen protein measurement by ProStatus prostate-specific antigen EQM Autodelfia Time-resolved fluoroimmunoassay system (Perkin-Elmer, Boston, MA). Ethical approval for blood sample collection for this research was obtained from the Leeds Teaching Hospitals NHS Trust ethics committee; all of the blood samples were collected after informed patient or volunteer consent.

The prostate cancer cell line LNCaP that expresses prostate-specific antigen was used as a positive control and was cultured in RPMI 1640 media (Invitrogen, Life Technologies, Inc., Paisley, England) supplemented with 10% (v/v) fetal calf serum (Harlan Sera-lab, Loughborough, England). All of the media contained 2 mmol/L glutamine.

To assess sensitivity of real-time RT-PCR for prostate-specific antigen mRNA detection, LNCaP cell line total RNA (0.001–100 ng) was serially diluted in diethyl pyrocarbonate-treated double distilled water and analyzed using real-time RT-PCR (see below). The sensitivity of real-time RT-PCR for prostate-specific antigen mRNA to detect prostate cancer cells (0.001–10,000) in blood (1 mL) was evaluated by adding known numbers of LNCaP cells to blood samples from healthy volunteers (cell spiking experiments).

Peripheral blood (5 mL) from healthy male volunteers (n = 5) was collected into EDTA-containing blood collection tubes (BD Vacutainer) and layered on to 5 mL of preformed gradient Polymorphprep (Axis Shield, Huntingdon, England). Cell separation was performed as per the manufacturer’s instructions and RNA extracted immediately from the mononuclear and polymorphonuclear cell fractions.

Total RNA was extracted from the LNCaP cell line, whole blood, and whole blood spiked with LNCaP cells using Ultraspec RNA isolation solution (Biogenesis) as described previously (21). Isolated RNA was washed in EtOH and dissolved in diethyl pyrocarbonate (Sigma, Dorset, England). RNA extraction from mononuclear and polymorphonuclear cells was performed using the RNeasy Midi kit (Qiagen, Crawley, England) as per the manufacturer’s instructions. Quantity of recovered RNA and its purity was measured by reading absorbance at 260 and 280 nm.

All of the steps were performed using sterile technique in separate rooms with designated areas for RNA extraction, RT-PCR, and sample analysis to reduce the risk of contamination to samples.

Polyadenylated mRNA was isolated from total RNA (5 μg) using magnetic Dynabeads Oligo (dT)₂₅ (Dynal Biotech, Bromborough, England) as described previously (22). Polyadenylated mRNA was then reverse transcribed to cDNA in a reaction volume of 20 μL, containing polyadenylated mRNA, random hexamer primer (250 ng, Invitrogen, Life Technologies, Inc.), 0.5 mmol/L deoxynucleoside triphosphates (Ultrapure dNTP Set, Amersham Biosciences, Buckinghamshire, England), 1 × First Strand Buffer (Invitrogen, Life Technologies, Inc.), 10 mmol/L DTT (Invitrogen, Life Technologies, Inc.), 28 units RNase inhibitor (RNA Guard, Amersham Pharmacia Biotech), 200 units of reverse transcriptase (SUPERSCRIPT II RT, Invitrogen, Life Technologies, Inc.), and distilled water (UltraPURE, Distilled Water, DNase, RNase Free, Invitrogen, Life Technologies, Inc.). Samples were then incubated at 42°C for 50 minutes and the reaction inactivated by heating at 70°C for 15 minutes. Samples were placed on ice until amplification by PCR.

Real-time PCR was performed using the ABI PRISM 7700 Sequence Detection System (Applied Biosystems, Warrington, England). The primers and TaqMan probe (Applied Biosystems) for the prostate-specific antigen assay were designed using the Primer Express software (version 1.5, Applied Biosystems), using the mRNA and genomic DNA sequences for prostate-specific antigen (GenBank accession numbers X07730 and M24543, respectively). The TaqMan probe for prostate-specific antigen was designed to anneal across exon 2/exon 3 junction. Prostate-specific antigen primer sequences were forward (exon 2) TCTGCGGCGGTGTTCTG and reverse (exon 3) GCCGACCCAGCAAGATCA to give a PCR product size of 68 bp. The prostate-specific antigen carboxyfluorescein-labeled probe sequence was CTGCCCACTGCATCAGGAACAAAAGC.

For each 96-well plate experiment LNCaP RNA was used as a positive control for amplification of prostate-specific antigen mRNA. The stably expressed endogenous control gene β₂-microglobulin was amplified as a control for quality and quantity of in-put RNA (23, 24). Published β₂-microglobulin primers and probe sequences were used (23): forward (exon 2) GAGTATGCCTGCCGTGTG and reverse (exon 4) AATCCAAATGCGGCATCT to give a PCR product size of 100 bp. The carboxyfluorescein-labeled probe sequence was CCTCCATGATGCTGCTTACATGTCTC.

The positive control for the β₂-microglobulin assay was RNA from a healthy volunteer. A negative control for each sample lacking reverse transcriptase was also included to confirm that amplification products were generated from cDNA rather than contaminating genomic DNA. Also, a triplicate water control lacking RNA in the PCR was included to control for any contaminating cDNA.

A 50-μL PCR contained 5 μL of cDNA (derived from 1 μg total RNA), primers (prostate-specific antigen forward 50 nmol, reverse 900 nmol; β₂-microglobulin forward 300 nmol, reverse 900 nmol), TaqMan probe (prostate-specific antigen 125 nmol; β₂-microglobulin 175 nmol), distilled water (UltraPURE, Distilled Water, DNase, RNase Free, Invitrogen, Life Technologies, Inc.), and 1 × TaqMan Universal PCR Master Mix (Applied Biosystems). The TaqMan Universal PCR Master Mix (Applied Biosystems) contained AmpliTaq Gold DNA polymerase, AmpErase uracil-N-glycosylase, deoxy-nucleoside triphosphates, MgCl₂, and Taqman buffer. The universal cycling parameters were 50°C for 2 minutes, 95°C for 10 minutes, and then 50 cycles of denaturation at 95°C for 15 seconds and annealing and extension at 60°C for 1 minute.

Each sample of unknown quantity was analyzed in triplicate for the target gene. Reaction mix (3 volumes) as described above containing primers and probe was aliquoted into microcentrifuge tubes and sample cDNA (3 volumes) added to each tube and mixed to give a final volume of 150 μL. This volume was divided into 3 by loading 50 μL into each of 3 wells of a 96-well Optical Reaction Plate (Applied Biosystems). The plate was sealed by placing Optical Caps (Micro Amp, Applied Biosystems) and placed in the ABI PRISM 7700 Sequence Detection System (Applied Biosystems) and the PCR program initiated as per the manufacturer’s instructions.

PCR for β₂-microglobulin was performed once; optimization experiments showed that PCR for β₂-microglobulin in triplicate was highly reproducible with a low intra-assay coefficient of variation (0.5%).

The efficiency of amplification of prostate-specific antigen and β₂-microglobulin were found to be equal. This enabled relative quantification to be performed using the comparative C_T method that uses an arithmetic formula (2^-ΔΔCT), which requires equal efficiency of the PCRs (Applied Biosystems). ΔC_T is the difference in the C_T values between the target prostate-specific antigen (average C_T used, as each sample was analyzed in triplicate) and the endogenous control β₂-microglobulin. ΔΔC_T is the difference between ΔC_T of a sample and the ΔC_T of a calibrator sample. The calibrator sample is a sample chosen against which other samples are compared. To compare levels of prostate-specific antigen mRNA between healthy volunteers and patients with prostate cancer, one of the healthy volunteer blood samples was chosen as the calibrator sample to which all of the other samples were compared.

Serum prostate-specific antigen protein level was measured using the ProStatus prostate-specific antigen EQM Autodelfia Time-resolved fluoroimmunoassay system (Perkin-Elmer). Blood samples were analyzed by the Department of Clinical Biochemistry, St. James’s University Hospital. The upper reference limit for a normal test was 4.0 μg/L.

A comparison of β₂-microglobulin mRNA levels and prostate-specific antigen mRNA levels among healthy volunteers, patients with localized prostate cancer, and patients with metastatic prostate cancer was performed using one-way ANOVA (SPSS version 11.0 for Windows 2000) and nonparametric Kruskal-Wallis test followed by post hoc Dunn’s multiple comparisons test (GraphPad Instat 3), respectively. Differences were considered significant if P < 0.05. Intra-assay variation for prostate-specific antigen mRNA expression (ANOVA) was expressed as the coefficient of variation (%). A comparison of the number of positive reactions in a triplicate reaction for prostate-specific antigen mRNA between the groups was performed using the exact χ² test (SAS, version 8.01). Survival analysis factored for whether a patient was RT-PCR positive or not for prostate-specific antigen mRNA was performed using log rank test and data visualized by plotting a Kaplan-Meier survival curve (SPSS version 11.0 for Windows 2000).

A comparison of β₂-microglobulin expression levels between healthy volunteers and patients with prostate cancer was made using all of the available C_T values for β₂-microglobulin (including results on follow-up blood samples). There was no significant difference in the mean C_T values of healthy volunteers, patients with localized prostate cancer, and patients with metastatic prostate cancer (P = 0.485; Fig. 1). There was also no effect of hormonal treatment on β₂-microglobulin expression (P = 0.91). The lack of significant variation in β₂-microglobulin expression between healthy volunteers and diseased patients and the lack of effect of hormonal treatment on β₂-microglobulin expression deemed it suitable for use as an endogenous control for the relative quantification of prostate-specific antigen mRNA in patients with prostate cancer.

Real-time RT-PCR reproducibly detected prostate-specific antigen transcripts in as little as 1 pg of total RNA extracted from the LNCaP cell line and β₂-microglobulin transcripts in 1 pg of total RNA extracted from 1 mL of blood taken from a healthy volunteer. In cell spiking experiments where decreasing numbers of LNCaP cells were added to 1 mL of whole blood from a healthy volunteer, real-time RT-PCR consistently detected 1 cell in a background of 10⁶ leukocytes.

To compare the reproducibility of each PCR performed in the triplicate PCR, the intra-assay variability of RT-PCR for prostate-specific antigen mRNA detection was calculated. All of the available patient data for blood samples that exhibited two or three reactions positive in a triplicate reaction was used. This revealed a low coefficient of variation (2.2%).

Real-time RT-PCR detected prostate-specific antigen mRNA in 14 of 21 blood samples from healthy volunteers (2 of 5 female volunteers and 12 of 16 male volunteers were positive). To compare levels of prostate-specific antigen mRNA between healthy volunteers and patients with prostate cancer, one of the healthy volunteer blood samples was chosen as the calibrator sample to which all of the other samples were compared. In the blood samples that were positive for prostate-specific antigen mRNA, the median relative prostate-specific antigen mRNA level was 0.97 (range, 0.08 to 2.35). Analysis of the mononuclear and polymorphonuclear cell fractions separated from buffy coat samples isolated from peripheral blood of healthy volunteers identified prostate-specific antigen expression by the mononuclear fraction in 5 of 5 blood samples. Relative prostate-specific antigen mRNA levels were similar to those observed previously for healthy controls (Table 1).

Analysis of blood samples from patients with metastatic prostate cancer identified 32 of 40 (80%) positive cases for prostate-specific antigen mRNA. The median relative prostate-specific antigen mRNA level was 7.08 (range, 1.02 to 1,540.31) -fold higher than in healthy volunteers. The absence of any prostate-specific antigen mRNA in 8 of 40 blood samples may reflect sample degradation in these patients or a genuine absence of circulating tumors cells.

Of the blood samples from patients with localized prostate cancer 10 of 27 (37%) were positive, with a median relative prostate-specific antigen mRNA level 1.62 (range, 1.02 to 3.02) times higher than in healthy volunteers. An overlap in relative prostate-specific antigen mRNA levels in blood samples from patients with prostate cancer and healthy volunteers was observed (Fig. 2). However, there was a significant difference in the relative prostate-specific antigen mRNA levels between the groups (P < 0.001; Kruskal-Wallis test). Post hoc Dunn’s multiple comparisons test revealed significantly higher levels of prostate-specific antigen mRNA in blood samples from patients with untreated newly diagnosed metastatic prostate cancer as compared with healthy volunteers (P < 0.01) and patients with localized prostate cancer (P < 0.001). Similarly, blood samples from a heterogeneous group of patients with metastatic prostate cancer that included patients receiving hormonal treatment intermittently but who at the time of blood sampling were off treatment (n = 6), patients who had developed androgen independent disease (n = 5), and 1 patient on hormonal therapy that was still responding to treatment also had significantly higher levels of prostate-specific antigen mRNA than healthy volunteers (P < 0.05) and patients with localized prostate cancer (P < 0.001). No difference in levels was found between blood samples from healthy volunteers and patients with localized prostate cancer (P > 0.05).

It was also noted that within a triplicate PCR, blood samples from patients with metastatic prostate cancer were significantly more likely to be positive for prostate-specific antigen mRNA in all three of the reactions; this was particularly true in samples from patients who had never received treatment (P < 0.0001, χ² test; Fig. 3). Similarly, healthy volunteers were more likely to be positive for one reaction out of a triplicate PCR (P < 0.0001, χ² test).

To identify patients with metastatic prostate cancer from other prostate cancer patients or healthy volunteers, the number of positive reactions in a triplicate PCR and relative prostate-specific antigen mRNA levels were examined together to establish the most appropriate cutoff level (Table 2). A positive test was defined as two reactions positive in a triplicate reaction in combination with a relative prostate-specific antigen mRNA level of >2.4 times that of healthy volunteers level or three reactions positive in a triplicate reaction regardless of relative prostate-specific antigen mRNA level provided the greatest overall sensitivity (68%) and specificity (95%) of detection. This definition was used for interpretation of all future data.

Examination of the relationship between serum prostate-specific antigen protein levels and relative prostate-specific antigen mRNA expression in blood samples from patients with metastatic prostate cancer revealed no correlation (Fig. 4), suggesting that the assays may reflect different characteristic features of prostate cancer, i.e., that circulating prostate-specific antigen mRNA levels reflects prostate cancer biology, whereas prostate-specific antigen protein levels may reflect tumor burden only.

Overall, the level of serum prostate-specific antigen protein was more frequently elevated than levels of prostate-specific antigen mRNA (above the normal range defined in this study) in blood samples from patients with newly diagnosed metastatic prostate cancer (28 of 28, 100% versus 19 of 28, 68%). This was also the case for patients with localized prostate cancer (27 of 27, 100% versus 2 of 27, 7%).

In a subset of patients with newly diagnosed untreated metastatic prostate cancer (n = 14) blood samples were taken at diagnosis and during hormonal therapy. In 10 of 14 samples prostate-specific antigen mRNA was detected at diagnosis. These levels declined in 8 of 10 patients during treatment. In these patients the level of serum prostate-specific antigen protein levels also dropped (Table 3). This was associated with a clinical response to treatment as evidenced by improvement in bone scan appearances in 3 patients, improvement in computed tomography scan appearance in 1 patient, and improvement in symptoms in the remaining 4 patients. In 3 patients the levels of relative prostate-specific antigen mRNA increased during treatment. For 2 of these patients (patient 12 and 14) the follow-up samples were taken during relapse after an initial response to treatment, and this was associated with rising serum prostate-specific antigen protein levels. In patient 12 this was lower than the level at diagnosis but was clearly rising. In 1 patient no response to treatment had been observed (patient 5); this patient also showed a rise in serum prostate-specific antigen protein level.

In a small cohort of patients with newly diagnosed metastatic prostate cancer (n = 28) blood samples for RT-PCR analysis were taken before the start of treatment. At the time of this report 15 deaths had occurred; 11 of 15 of these deaths were due to progressive prostate cancer. Four deaths were due to other causes. Comparison of overall survival for those patients with blood samples that were positive by RT-PCR for prostate-specific antigen mRNA as defined in this study to those who were negative revealed no difference (log rank, P = 0.3; Fig. 5). Examining different cutoffs in relative prostate-specific antigen mRNA levels did not predict for survival.

Association with progression-free survival was not performed, because routine clinical evaluation of relapse was not sufficiently consistent enough to provide accurate data.

In this report we describe a highly sensitive and specific quantitative RT-PCR for prostate-specific antigen mRNA that allows the objective discrimination of blood samples from patients with metastatic prostate cancer from those with localized disease and healthy volunteers. An overall detection sensitivity of 68% was attained with a specificity of 95%, when a positive test was defined as two positive reactions in a triplicate reaction in combination with a relative prostate-specific antigen mRNA level (defined by relative C_T value) that was 2.4-fold higher than that of healthy volunteers or if three reactions were positive in a triplicate reaction regardless of the relative prostate-specific antigen mRNA level. For newly diagnosed patients with metastatic prostate cancer a detection sensitivity of 70% was achieved using this definition. However, in 8 of 40 (20%) of these patients no prostate-specific antigen transcripts were detected. This may reflect sampling error or degradation of transcripts where levels may have been low despite PCR for β₂-microglobulin suggesting that good quality RNA was analyzed.

Quantification of prostate-specific antigen mRNA expression in blood samples not only enables identification of patients with disseminated disease but also allows monitoring of treatment effect on disseminating disease as demonstrated in the group of newly diagnosed patients with metastatic prostate cancer. RT-PCR for prostate-specific antigen mRNA detection in blood may be exploited to monitor the effect of conventional therapies in patients with metastatic disease and for the initial evaluation of novel therapies for prostate cancer where currently tools for objective assessment are lacking (25).

In patients with localized prostate cancer, levels of prostate-specific antigen mRNA expression were low and no different from those identified in blood samples from healthy volunteers. This result is in contrast to previous studies using conventional nonquantitative RT-PCR or quantitative real-time RT-PCR (26, 27). Previous studies have not compared levels in patient samples with those from healthy volunteers. The discordant results most likely reflect the increased discriminatory power of the real-time quantitative method and definition of what constitutes a positive result. From the results of the current study where detailed examination of prostate-specific antigen transcript levels in blood samples from healthy volunteers was performed, it is clear that some apparently positive patient samples in other studies may have been “false” positive. The effect of this is probably reflected in the conflicting results from many of the published studies to date (5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15). Using the highly sensitive quantitative real-time RT-PCR we confirm the presence of low-level expression of prostate-specific antigen by monocytes in blood samples from healthy volunteers. The illegitimate transcription of tissue-specific genes in blood from healthy volunteers has been described previously, for example cytokeratin 19 (CK19) expression by mononuclear cells (28). The results from the current study confirm the more recent findings of illegitimate expression of prostate-specific antigen mRNA by the CD34⁺ fraction of mononuclear cells (29). This latter study also noted illegitimate transcription of many other tissue-specific genes including prostate-specific membrane antigen (PSMA), CK19, and CK20 and carcinoembryonic antigen. Compensation for the low-level signal due to illegitimate transcription of the target gene is important if the true clinical utility of RT-PCR is to be realized. Others have sought to achieve this by either selecting out the monocyte population (CD34⁺ cells) before RNA extraction or RT-PCR (29), although this risks losing circulating tumor cells during the isolation process and requires additional sample manipulation. Alternatively, some have suggested reducing the amount of total RNA or cDNA analyzed by the RT-PCR (28, 29), which may improve specificity but will reduce the overall sensitivity of detection of circulating tumor cells. Therefore, we believe that establishing a robust definition for contamination of a blood sample with a prostate cancer cell based on the prostate-specific antigen mRNA level compared with that in blood samples from healthy controls is the most reliable and sensitive method to detect clinically relevant disease.

Elevated prostate-specific antigen mRNA in blood samples from patients with metastatic prostate cancer is consistent with the hypothesis that these patients have circulating tumor cells that might go on to develop metastases. Consequently the detection of elevated levels of prostate-specific antigen mRNA might be more informative about the potential of a prostate cancer cell to metastasise and inform on overall outcome than an elevated serum prostate-specific antigen protein level alone, which may only reflect total body tumor bulk. It is encouraging that no correlation between these two tests was seen suggesting that both may provide useful but different information. It is recognized that prostate-specific antigen protein level may be of predictive value when circulating levels fall significantly during treatment and fall into the normal range (25). The prognostic significance of a positive RT-PCR test for prostate-specific antigen mRNA was evaluated recently in patients with androgen-independent disease and was found to predict for shorter survival (30). We also examined the effect of a positive RT-PCR test for prostate-specific antigen mRNA on prognosis in a group of patients with newly diagnosed metastatic prostate cancer but found that it did not predict for survival; this evaluation was limited by small sample size. A larger prospective clinical study evaluating the role of RT-PCR is needed to definitively address this question.

In conclusion, in this study prostate-specific antigen mRNA has been detected in peripheral blood samples from healthy volunteers consistent with previous studies. However, using quantitative real-time RT-PCR it is possible to discriminate between this illegitimate expression of prostate-specific antigen mRNA and the presence of prostate-specific antigen mRNA in peripheral blood from patients with metastatic disease. Using this quantitative assay it is now necessary to evaluate the clinical significance of prostate-specific antigen mRNA in peripheral blood from patients with metastatic prostate cancer in a large prospective clinical outcome study to determine the utility of this sensitive technique.

The authors wish to acknowledge the contribution of the following nurses and laboratory personnel: Elizabeth Hudson, Paul Berry, Andrea Simpson, and Paul Evans.