| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
a Address correspondence to this author, at: Department of Urology, Mail Stop 356510, University of Washington, Seattle, WA 98195. Fax 206-543-1146; e-mail ecorey{at}u.washington.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Key Words: indexing terms: cancer • metastasis • peripheral blood mononuclear cells
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Gene-amplification techniques, with their high sensitivity, can be used
to address this need. Since 1987, reverse transcriptase–polymerase
chain reaction (RT-PCR) methodology has been used to detect circulating
cells shed by solid tumors, utilizing, e.g., tyrosinase for melanoma
(1); tyrosine hydroxylase for neuroblastoma
(2); 
-fetoprotein for hepatoma (3); and
cytokeratin for breast cancer
(4).1
This broad spectrum demonstrates the diagnostic potential of
the RT-PCR method. The premise of this approach is that the cells to be
detected express mRNA that is not found in other cells within the
environment being tested. For detecting prostate cells in circulation,
the target transcripts of choice are currently prostate-specific
antigen (PSA) and prostate-specific membrane antigen
(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15). Despite the promise of this method, however, it
is not yet clear whether RT-PCR will prove to be of clinical value in
staging prostate cancer. Results to date are mixed
(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27). The efficiency of circulating tumor cells in
causing metastasis is also controversial (28)(29)(30)(31)(32)(33). Better
methods and protocols, along with more-comprehensive studies with
longer follow-up, are needed to answer these questions.
Here we report an enhanced RT-PCR protocol for detecting prostate cells in circulation by targeting PSA sequences. We have designed and evaluated endogenous and exogenous internal and external controls to ensure uniform performance of the tests. The detection limit of the protocol is ~1 PSA-positive LNCaP cell per 108 peripheral blood mononuclear cells (PBMC), or 5 copies of PSA cDNA, with no signal from negative-control samples.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Collection of blood samples and isolation of mononuclear
cells.
Samples of venous blood (8 mL) were drawn into Vacutainer
Tube CPT cell-preparation tubes (Becton Dickinson, Franklin Lakes, NJ)
and processed within 2 h. After the samples were centrifuged at
1500–1800g for 30 min at room temperature, the resulting
PBMC layers were gently resuspended in plasma and transferred to 50-mL
tubes. The volume was brought to 40 mL with Dulbecco's
phosphate-buffered saline, and 10-µL aliquots were removed and
diluted with 40 µL of 30 mL/L acetic acid for cell counts. The
remaining cells were centrifuged at 300g for 20 min at
4 °C and the cell pellets were resuspended in STAT-60 (Tel-Test
"B", Friendswood, TX) for RNA isolation.
Collection of bone marrow samples and isolation of
mononuclear cells.
Bone marrow samples (10 mL) were drawn into
20-mL syringes containing 10 mL of 60 g/L sodium citrate solution
(Pharmaceutical Services, University of Washington, Seattle, WA). The
mixtures were transferred to 50-mL tubes, underlaid with 15 mL of
Histopaque-1077 (Sigma, St. Louis, MO), and centrifuged at 400gfor 30 min at room temperature. The resulting opaque layers
containing the mononuclear cells were then transferred to new tubes and
brought to 40 mL volume with phosphate-buffered saline. The cells were
isolated for counting as described for blood samples. We processed all
of the isolated PBMC for RNA extraction to maximize the chance of
detecting prostate cells.
procedures
RNA isolation.
Total RNA was isolated from PBMC by using
STAT-60 Total RNA Isolation Reagent (Tel-Test "B") according to the
manufacturer's recommendations, except that we used 1 mL of STAT-60
per 5 x 106 PBMC, rather than 1 mL per 10 x
106 PBMC. The isolated RNA was dissolved in 15 µL of
"Molecular-Biology-grade" water (5prime-3prime, Boulder, CO) in
siliconized, RNase- and DNase-free microcentrifuge tubes. The RNA
concentration was determined by measuring absorbance at 260 nm. A
5-µg aliquot of RNA was used for reverse transcription (RT) and the
remainder was stored under ethanol at -80 °C.
Reverse transcription.
RT was performed in a volume of
20 µL, as follows: 5 µg of total RNA and 1 µL (0.05 µg) of
Random Hexamers (Perkin-Elmer, Branchburg, NJ) were brought up to 10
µL with Molecular-Biology-grade water, heated to 70 °C for 10 min,
and then cooled immediately on ice. A master mixture was prepared, 10
µL for each sample, as follows: 4 µL of 5x First Strand Buffer
(i.e., 250 mmol/L Tris-HCl, pH 8.3 at room temperature, 375 mmol/L KCl,
and 15 mmol/L MgCl2); 2 µL of 0.1 mol/L dithiothreitol; 1
µL of a mixture of dATP, dGTP, dCTP, and dTTP, each at 2.5 mmol/L; 1
µL of RNase Inhibitor (10 units; Perkin-Elmer); 1 µL of reverse
transcriptase (200 units, Superscript II Reverse Transcriptase;
Gibco-BRL Life Technologies, Grand Island, MD), and 1 µL of
H2O. The master mixture was added to each sample tube
(containing RNA and primers) and heated for 5 min at 25 °C, followed
by 1 h at 42 °C. The reverse transcriptase was deactivated by
heating at 99 °C for 5 min.
ß2-Microglobulin (MIC) PCR.
MIC mRNA
(43) was used as an endogenous external control for RT.
PCR with MIC primers (28-mers; Integrated DNA Technologies, Coralville,
IA) yields a 550-bp fragment spanning three exons at positions
148–401, 1018–1045, and 2290–2560 of the MIC gene (GenBank M17987).
The specificity of the MIC primers was verified with the GCG program
Version 8.1 UNIX (Genetics Computer Group, Madison, WI). Their
structure was as follows:
5' MIC primer CACGTCATCCAGCAGAGAATGGAAA-GTC 3' MIC primer TGACCAAGATGTTGATGTTGGATAAGAG
PCR with these primers was performed in a 50-µL volume, with 1 µL of prepared cDNA and 5 pmol of each primer per reaction. A master mixture prepared for a set of simultaneous PCR reactions consisted of 5 µL of 10x PCR Buffer (200 mmol/L Tris, pH 8.3, plus 500 mmol/L KCl), 2 µL of 50 mmol/L MgCl2 (final concentration 2 mmol/L), and 4 µL of a mixture containing 0.5 mmol/L of each dNTP (final concentration 40 µmol/L) per sample. To ensure specificity of the MIC PCR, we used hot-start conditions: For each sample we mixed 0.5 µL of Taq DNA Polymerase (2.5 units; Gibco-BRL) with 0.5 µL of TAQSTART antibody (Clontech Labs., Palo Alto, CA) and incubated this for 5 min at room temperature before adding it to the master mixture. Cycling conditions (Omnigene Thermal Cycler, Hybaid, Middlesex, UK) were: 80 °C for 3 min, 1 cycle; 94 °C for 5 s and 69 °C for 1 min, 25 cycles; and 72 °C for 7 min, 1 cycle. Reaction products were analyzed by electrophoresis in 1.5% agarose gels with 50 mmol/L Tris-borate buffer containing 1 mmol/L EDTA, and were stained with ethidium bromide.
PSA PCR.
The PSA oligonucleotides (Integrated DNA
Technologies) were chosen from regions exhibiting high sequence
divergence between PSA cDNA and the closely related human glandular
kallikrein cDNA (hK2; GenBank M18157), which also is expressed by
prostate tissue.
5' PSA primer TTGTGGCCTCTCGTGGCAGGGCAGT
3' PSA primer TGGTCACCTTCTGAGGGTGAACTTGC
PSA cDNA and hK2 cDNA differ at 10 positions within the sequence of the 5' primer and 9 positions within the 3' primer. Because the primer sequences are from different exons of PSA mRNA (5' PSA, exon 2; 3' PSA, exon 4), the researcher can easily determine the presence of genomic DNA contaminants, which would yield much larger products. The 460-bp PCR fragment obtained with these primers has the same sequence as positions 110–569 of PSA cDNA (GenBank M21895).
Specificity of the primers was verified with the GCG UNIX program (see above). PCR was performed in 50-µL volumes as described for MIC PCR, except that 5 µL of prepared cDNA and 5 pmol of each primer were used and the cycling conditions were 80 °C for 3 min, 1 cycle; 94 °C for 5 s; 69 °C for 1 min, 40 cycles; 72 °C per 7 min, 1 cycle.
Restriction enzyme digestion.
To 13 µL of the PSA PCR
samples were added 1.5 µL of 10x ClaI enzyme reaction
buffer (500 mmol/L Tris-HCl, pH 8, and 100 mmol/L MgCl2)
and 0.5 µL of ClaI enzyme (both from Gibco- BRL). The
reaction mixture was incubated for 1 h at 37 °C, and the
products were analyzed by agarose gel electrophoresis.
Cloning and sequencing
PCR products. PSA PCR
products were cloned directly with use of a TA Cloning Kit (Invitrogen,
San Diego, CA). Plasmids were transformed into electro-competent
Escherichia coli JM109, purified with a Wizard Plasmid
Preparation Kit (Promega, Madison, WI), screened by restriction
analysis, and sequenced.
determination of detection limit
The detection limit of our protocol was tested in three ways.
First, we determined the detection limit of the PCR step by performing
PCR with serial dilutions of PSA cDNA cloned into pGEM-3Z (gift of
ke Lundwall, Malmo, Sweden), 0 to 160 copies per
reaction. Second, we added LNCaP cells total RNA or LNCaP cells to
total PBMC RNA or PBMC, respectively, at a ratio of 1 LNCaP cell (or
RNA equivalent) per 104, 105,
106, 107, or 108 PBMC (or RNA
equivalent) and assayed by RT-PCR to determine how few cells could be
detected. Third, to test the detection limit of the entire procedure,
we added 2–20 LNCaP cells to 8 mL of PSA-negative blood from female
donors. Confluent LNCaP cells were harvested, washed, and counted with
a hematocytometer; we then diluted 106 cells resuspended in
1 mL of RPMI 1640 in 10-fold serial dilutions. For the
105/mL dilutions of LNCaP cells, the cells were counted by
standard procedures with a hematocytometer; for lower dilutions
(104, 103, and 102/mL), we counted
the cells in 10 or 20 µL of the cell suspension pipetted onto a
microscope slide with a grid. We then pipetted 20–200-µL aliquots of
the 102/mL LNCaP cell preparation (containing 2, 5, or 20
cells) into PSA-negative blood samples and processed the cells as
described above. The expected frequency of occurrence of various
numbers of cells at extreme dilutions was calculated according to the
binomial distribution since the number of cells observed is very small
relative to the number of cells in the pool (1 x 107 to 2
x 107).
exogenous internal competitive dna control (ic-psa)
Construction.
PSA cDNA excised by EcoRI from
pGEM-3Z was recloned into the unique EcoRI site of the
plasmid pLOT731 (unlike pGEM-3Z, pLOT731 contains no SacI
site). The resulting plasmid was linearized at position 420 of the PSA
cDNA by SacI digestion. After treatment with phosphatase
(Boehringer Mannheim, Indianapolis, IN), an unrelated PCR fragment 258
bp long was inserted into the SacI site. The insert was
prepared by PCR amplification of FAU cDNA positions 164–416
(GenBank X65923) with primers containing SacI restriction
sites. This PCR-amplified fragment was digested with SacI,
gel-purified, and ligated into the SacI-digested pLOT731 PSA
clone. The plasmid was transformed into E. coli JM109.
Bacterial colonies containing the desired extended PSA sequences were
identified by PCR. Desired colonies yielded a PCR band of 720 bp,
representing IC-PSA. This IC-PSA cDNA was excised with EcoRI
and cloned back into pGEM-3Z for in vitro transcription.
PCR performance evaluation.
Plasmids (pGEM-3Z)
containing either PSA cDNA or IC-PSA cDNA were serially diluted and
combined for use in PCR. We used equal amounts of each cDNA in the
range 0.00001–1 pg (5- 500 000 copies) of each plasmid DNA. In
addition, we used three different amounts of IC-PSA plasmid DNA
(0.0001, 0.001, and 0.01 pg, equivalent to 50, 500, and 5000 copies,
respectively) in combination with 0.00001–1 pg (5–500 000 copies) of
PSA cDNA/pGEM-3Z.
In vitro RNA synthesis.
IC-PSA cDNA in pGEM-3Z was used
for RNA synthesis. The IC-PSA cDNA orientation in pGEM-3Z was
determined by digestion with BamHI and SacI. An
SP6 Megacript Kit (Ambion, Austin, CA) was used according to the
manufacturer's recommendations to synthesize the exogenous internal
competitive IC-PSA RNA. The plasmid was linearized at position 635 of
the PSA sequence with ApaI; the resulting transcript was
~780 bp long. The RNA was used in RT-PCR to verify its identity.
Identical reactions, but with reverse transcriptase omitted, were used
to monitor contamination by plasmid-derived IC-PSA cDNA.
RT-PCR performance evaluation.
Various concentrations of
IC-PSA RNA were added to 5 µg of PSA-negative PBMC RNA that had been
adulterated with 5 x 10-7 µg of RNA
from LNCaP cells. These mixtures were then subjected to RT-PCR as
described above.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
To determine the detection limit of the optimized PCR protocol, we
performed PCR on serial dilutions of pGEM-3Z with PSA cDNA, yielding
0–160 copies of plasmid per reaction. Results are shown in Fig. 1
. Specificity of the primers for PSA was checked by PCR with an
hK2-containing plasmid (gift of D. Tindall, Rochester, NY); no band was
observed, even with high concentrations of hK2 cDNA (data not shown).
|
For evaluating and optimizing the RT portion of the procedure, we
performed RT with serial dilutions of LNCaP RNA added to PBMC RNA to
yield ratios of 1 prostate cancer cell to 104,
105, 106, 107, or 108
PBMC. We tested the use of 1, 2, or 5 µg of RNA per reaction and 2,
3, and 5 µL of RT product per PCR. The conditions exhibiting the
lowest limit of detection and best reproducibility for detection of the
PSA message were 5 µg of RNA per RT reaction and 5 µL of RT product
per PCR. PCR with MIC primers (the endogenous external control) was
used as a positive control for assessing RT performance and the
presence of amplifiable messages. Using these conditions, we could
detect 1 LNCaP cell per 108 PBMC, as shown in Fig. 2
. We obtained a positive signal from 1 LNCaP cell in a
background of 108 PBMC in five out of five experiments
performed with various LNCaP- and PBMC-RNA preparations over ~6
months.
|
We optimized the conditions for RNA extraction by adding dilutions of LNCaP cells directly into PBMC in the same ratios as above and extracting these mixtures. We found that following the manufacturer's recommendation to use 1 mL of STAT-60 per 10 x 106 cells resulted in persistent DNA contamination, whereas using 1 mL of STAT-60 per 5 x 106 cells alleviated the problem. RT-PCR was performed as described above. The detection limit of this cell-addition procedure was also 1 LNCaP cell per 108 PBMC (data not shown).
The limit of detection for the whole method depends on sample
collection, blood handling, isolation of cells, and RNA extraction, as
well as the RT-PCR procedure itself. For example, in our experience,
blood collection into CPT tubes is superior to Ficoll methods for
isolation of PBMC and circulating prostate cells. Final testing of the
whole procedure was performed with 8-mL samples of blood supplemented
directly with dilutions of LNCaP cells in RPMI 1640; the desired
numbers of cells were added in volumes of at least 20 µL to CPT tubes
containing blood. The average number of cells observed in 10 µL of a
theoretical dilution of 104 cells per 1 mL was
98.75 ± 10.62. We repeated these experiments over ~1 year with
different preparations of LNCaP cells to establish our limit of
detection, i.e., 2 LNCaP cells in 8 mL of peripheral blood (Fig. 3
). At this high dilution, the tube-to-tube variation in cell
content is large relative to the mean; therefore, we used the binomial
distribution, which showed that ~59.4% of the tubes would be
expected to have at least 2 cells, while 13.5% would be expected to
have none. To ascertain whether the actual cell delivery matched
theory, we performed cell counts with cell dilutions representing
105, 104, 103, and 102
LNCaP cells per milliliter. At a nominal dilution of 2 cells per 20
µL (100/mL), the counts followed a gaussian distribution and yielded
an average of 1.51 ± 0.99 cells per aliquot (n = 37). In
particular, whereas the theoretical expectation was that 59.4% of the
samples would contain 2 or more cells, the experimental value was
51.4%. A review of 12 RT-PCR experiments performed with the goal of
adding 2 LNCaP cells to 8 mL of blood showed that 7 (58.3%) of the
experiments yielded the PSA PCR product—in close agreement with the
theoretical expectation that 59.4% of the supplemented blood samples
would contain 2 or more cells.
|
Samples of peripheral blood (n = 40) and bone marrow (n = 30) from negative donors (females or young males) were used as a negative controls. None of the samples exhibited the 460-bp band for PSA mRNA. Preliminary data obtained from clinical trials by this procedure were presented at the 91st Annual American Urological Association Meeting (22)(23).
evaluation of controls
We constructed the IC-PSA cDNA in pGEM-3Z for use in control
procedures to evaluate RT-PCR performance. The IC-PSA plasmid contains
a PSA cDNA sequence into which an unrelated 258-bp sequence has been
inserted, and the IC-PSA cDNA can be amplified by the
same primers as the normal PSA cDNA sequence. We performed four
different experiments to test for simultaneous detection of the two
cDNAs. First, we amplified equimolar mixtures of PSA cDNA/pGEM-3Z and
IC-PSA cDNA/pGEM-3Z in the range 0–500 000 copies (Fig. 4
). Electrophoresis on 1.5% agarose gels easily
discriminated the signals from these two sources. PSA cDNA
yielded the 460-bp product, whereas the IC-PSA cDNA
yielded a 720-bp product.
|
Second, because the amount of PSA message in clinical samples may vary,
we tested the effect of three quantities of IC-PSA cDNA/pGEM-3Z
(0.0001, 0.001, and 0.01 pg, or 50, 500, and
5000 copies) on PCR performance with PSA cDNA/pGEM-3Z in
the range 0.0001–1 pg, or 5–500 000 copies. Because the two
amplifications are competitive, the ratio of quantities within which
both signals can be detected is limited. Within a factor of 10 in
either direction, both bands were detected; at ratios outside this
range, however, the more numerous species dominated and the other
species was not observed. For example, Fig. 5
illustrates the use of 0.0001 pg (50 copies) of IC-PSA
plasmid with PSA plasmid quantities of 0.00001–1 pg (5–500 000
copies). Under these conditions we could detect both IC-PSA cDNA and
PSA cDNA in the lowest range in which the PSA cDNA itself is
detected (0.00001 pg or 5 copies of PSA cDNA/pGEM-3Z).
|
Figure 6
depicts the use of the IC-PSA control in PCR with clinical
samples. Six clinical samples were examined with and without the IC-PSA
control. The 720-bp band in the control lanes for samples 2, 3, and 4
indicates a proper performance of PCR, and these samples are clearly
identified as true-negative samples. For sample 5, no amplification
product was detected in either lane (lanes 5 and 5A),
indicating that this sample is not a negative sample but rather a case of PCR failure. Another sample from this patient
must be examined.
|
To explore the possibility of using IC-PSA at the RNA level to control
for performance of RT, we prepared IC-PSA RNA by in
vitro transcription. The RNA species obtained was ~780 nucleotides,
yielding a DNA fragment of 720 bp by RT-PCR. A control experiment
omitting the reverse transcriptase did not yield the 720-bp band. To
check the recovery of IC-PSA RNA, we added various amounts of the RNA
into aliquots of PBMC that had been adulterated with LNCaP cells,
1:107. The signal from the IC-PSA RNA was detected when the
IC-PSA RNA was present at 0.01–10 pg, whereas both IC-PSA and PSA
signals were detected for IC-PSA RNA at 0.01–1 pg (Fig. 7
). However, because we found that the detection limit of IC-PSA
RNA was highly variable, we adopted the approach of using an
endogenous, external control (MIC) for RT performance and RNA
integrity; this control procedure yields an extremely low rate of false
negatives.
|
An additional control procedure is used only with PSA cDNA-positive
samples. The product is digested by the restriction enzyme
ClaI, which digests the product of PSA cDNA amplification to
yield bands of 220 and 240 bp. Fig. 8
shows the results of ClaI digestion of the PSA and
IC-PSA PCR products, as well as a mixture of the two that had been
mixed with undigested PSA cDNA after digestion of the pair
of products to demonstrate that the undigested band can be easily
distinguished from all of the reaction products.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Because of its sensitivity and selectivity, RT-PCR offers significant improvements in the detection of circulating prostate epithelial cells. Moreover, the detection limit of RT-PCR is better than that of any other methods in current use, e.g., flow cytometry or immunohistochemistry. There is therefore considerable interest in using RT-PCR to detect prostate cells in peripheral blood and bone marrow samples from patients with prostate cancer (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32). The bone marrow is of particular interest because of the tendency of prostate cancer to metastasize to bone. In the case of breast cancer, which resembles prostate cancer in being hormone-responsive, there is a documented relation between the presence of breast cancer cells in bone marrow and metastasis (39)(40)(41)(42). Although RT-PCR alone cannot determine whether the PSA cDNA-positive cells detected have malignant properties, correlations with pathology data (6) suggest a possible link between the presence of such cells in circulation and occult metastatic spread of the prostate cancer. Before 1996, the detection limits published for RT-PCR of PSA mRNA did not surpass 1 LNCaP cell in a background of 106 negative cells, as determined by use of a combination of RT-PCR with Southern transfer (6), signal amplification by digoxigenin (8), biotin–streptavidin systems in post-PCR ELISA (14), or nested RT-PCR (7). Recent abstracts have reported detection of 1 LNCaP cell in 107 PBMC (Marakovskiy A, Halpert G, Stein B, Hixson DC. Improved RT-PCR detection of circulating tumor cells in patients with prostate cancer, Abstract at 87th American Association for Cancer Research Annual Meeting, 1996) and 108 PBMC (16)(17). At the 91th American Urological Association Annual Meeting, O'Hara et al. reported (unpublished data) that they detected PSA signal in control samples—7 of 90 men (7.8%), 4 of 44 women (9.1%), and 9 of 54 (16.7%) patients with biopsy-proven benign prostatic hyperplasia—by using the procedure described elsewhere at that meeting (16)(17). They hypothesized that basal amounts of PSA mRNA expression might be responsible for these false positives. However, the agreement between this "false-positive" RT-PCR rate in biopsy-negative patients (16.7%) and the known false-negative rate of biopsies (15–20%) is highly suggestive. That is, these results may not be false positives but rather indicate cases in which RT-PCR has surpassed biopsy as a means of early diagnosis of prostate cancer. Follow-up studies on the patients giving false-positive results in the RT-PCR assay are desirable.
With clinical utility in mind, we set out to develop an RT-PCR protocol that would be highly sensitive and easy to use. Our protocol has a detection limit of 1 LNCaP cell in 108 negative cells, with no false-positive results. Using oligonucleotides 25 and 26 bases long enables us to use a faster, two-step PCR, and hot-start conditions contribute to the high specificity of amplification.
Use of RT-PCR in the clinical setting requires proper controls to assure the validity of results. There are two ways of classifying RT-PCR controls: endogenous or exogenous, and internal or external; in our protocols, the control procedures to be performed with every run are an endogenous external control and an exogenous internal control. The first control is endogenous because it involves amplification of MIC, a species naturally present in the sample; it is external because the amplification is performed in a different reaction vessel. The second control involves addition of the artificially created—exogenous—IC-PSA cDNA to the reaction tube in which the PSA PCR is taking place; this is therefore an internal control procedure for the reaction. The combination of these two procedures ensures the integrity of both the sample and the amplification procedures and virtually eliminates the effect of tube-to-tube variability on the results of the test. In addition to these controls, a detection-limit control containing ~5 copies of PSA cDNA is also run side-by-side with the target samples, and a reaction without cDNA (water control) is run to monitor reagent contamination. Finally, restriction analysis by Cla I is used to confirm the identity of the product in PSA cDNA-positive samples.
The experiments with IC-PSA DNA showed that one could detect both
target and internal control sequences when the concentrations of the
two did not differ by more than 10-fold. Outside this range, the
more-numerous species dominated the reaction and prevented detection of
the other species. We therefore decided that 0.0001 pg (50 copies) of
the IC-PSA plasmid was the optimal amount to add to the PCR mixture for
use with clinical samples (Fig. 6
). This permits control of PSA PCR
performance at the lowest limit of detection. Higher concentrations of
target PSA cDNA dominate the reaction, preventing efficient
amplification of IC-PSA cDNA. However, in the lower limit of detection
range, the presence of the 720-bp IC-PSA band verifies the PCR
performance and greatly reduces the probability that an apparent
negative result is in fact a false negative.
Although IC-PSA was constructed for use as a control for both the DNA and RNA aspects of the reaction, we found that RT with the IC-PSA transcript was not sufficiently reproducible for use as a clinical control, possibly because of degradation of uncapped RNA, or loss through interaction with the container's surface at the low concentrations needed for this reaction. Trouble-shooting this control procedure will require additional effort. We therefore settled on the MIC message as a noncompetitive endogenous control for the RT phase of the reaction. The message transcribed from this housekeeping gene is known to be present at approximately constant amounts in all samples we are likely to encounter (43). This type of external control also provides information about the integrity of the RNA preparation. The choice of a gene for use as the endogenous external control is an important issue: The commonly used housekeeping genes glyceraldehyde-3-phosphate dehydrogenase and ß-actin both have the disadvantage that matching processed pseudogenes are known to exist in the human genome. If primers can anneal to a pseudogene, PCR can potentially amplify genomic DNA contamination in addition to the target cDNA, and the two products would be indistinguishable in size. This is a serious problem, because amplification of the pseudogene present in DNA contamination of the RNA preparation would lead to a false assurance that the RT was successful, when in fact the RT may have failed. One must therefore check for the presence of a pseudogene before settling on a particular housekeeping gene to use as a positive RT control.
To eliminate false-positive results, we further tested each
PSA-positive sample by digestion with ClaI.
Digestion of the 460-bp PSA product by this enzyme yields bands of 220
and 240 bp. The hK2 product, on the other hand, is not digested by this
enzyme. This additional step allows us to distinguish a true
PSA-amplification event from artifacts caused by mispriming of messages
of similar size—and presents a simple alternative to the use of
Southern transfer or DNA sequencing procedures to confirm the identity
of the amplification products. In the 1100 samples we have thus far
examined with this procedure, <1% have yielded 460-bp products that
were not digestible by ClaI. In samples that do not exhibit
the 460-bp band, however, we have occasionally seen a band of
~430–450 bp that is not digestible by ClaI. Careful
examination of electrophoresis results is therefore necessary. Despite
the very low frequency of occurrence of false positives, we have
performed ClaI digestion for all samples exhibiting bands of
a size similar to that of the expected PSA product. The IC-PSA cDNA
does not interfere with the interpretation of the digestion result, and
in fact represents a positive control for ClaI performance
(Fig. 8
).
Improvements in RT-PCR procedures leading to a lower detection limit are of potential clinical significance for at least two reasons. First, whereas a typical blood aliquot provides ~107 cells, a bone-marrow sample can easily provide 108 cells, which improves the desired detection limit by an order of magnitude. Second, the procedures described rely on the cell-line LNCaP as a source of PSA-positive cells; however, prostate-cancer cells express variable amounts of PSA mRNA (44). This is not to say that there is no lower bound to the detection limit desired. Smith et al. have detected PSA message in many cell lines that are not of prostate origin, as well as in PBMC of control subjects (11). Such ultrasensitive methods as the ones they have developed are not suitable for prognostic procedures in prostate cancer because a positive result, although it may in fact indicate the presence of PSA messages outside the prostate, has little or no pathological significance (45). In developing diagnostic assays incorporating RT-PCR, one must strike a balance between specificity and detection limit.
In our view, attempts to quantify the PSA-PCR results would complicate the procedure without adding benefits. True quantification of results would require a very detailed examination of the internal RNA controls; because of the wide variation in amounts of PSA mRNA present in prostate cancer cells, however, it is not clear that any improvements in clinical significance would result. Even if we could quantify precisely the number of PSA mRNA molecules in the sample, we would still not know whether the number obtained indicated few cells, each containing many PSA messages, or many cells that contained few messages.
Given that the degree of correlation between the presence of PSA-positive cells in blood or bone marrow and patient prognosis is not yet known, it is too early to assess the significance of the improved detection limit and specificity obtained with these procedures. Clinical trials with the improved RT-PCR protocol are currently underway, designed to evaluate the usefulness of PSA RT-PCR in peripheral blood samples and bone marrow specimens in staging prostate cancer before radical prostatectomy and as a prognostic indicator after prostatectomy (22)(23). After more-extensive evaluation with clinical samples, we expect to include IC-PSA as an internal control in the next series of clinical trials.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
1 Nonstandard abbreviations: PSA, prostate-specific antigen; RT, reverse transcription; PBMC, peripheral blood mononuclear cells; LNCaP, prostate-cancer cell line; MIC, ß2-microglobulin; IC-PSA, internal control for PSA. 
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  D. Schamhart, J. Swinnen, K.-H. Kurth, A. Westerhof, R. Kusters, H. Borchers, and C. Sternberg Numeric Definition of the Clinical Performance of the Nested Reverse Transcription-PCR for Detection of Hematogenous Epithelial Cells and Correction for Specific mRNA of Non-Target Cell Origin as Evaluated for Prostate Cancer Cells Clin. Chem., September 1, 2003; 49(9): 1458 - 1466. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D Grammatopoulos, Y Elliott, S C Smith, I Brown, R J Grieve, E W Hillhouse, M A Levine, and M D Ringel Measurement of thyroglobulin mRNA in peripheral blood as an adjunctive test for monitoring thyroid cancer Mol. Pathol., June 1, 2003; 56(3): 162 - 166. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Ylikoski, K. Pettersson, J. Nurmi, K. Irjala, M. Karp, H. Lilja, T. Lovgren, and M. Nurmi Simultaneous Quantification of Prostate-specific Antigen and Human Glandular Kallikrein 2 mRNA in Blood Samples from Patients with Prostate Cancer and Benign Disease Clin. Chem., August 1, 2002; 48(8): 1265 - 1271. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Corey, J. E. Quinn, M. J. Emond, K. R. Buhler, L. G. Brown, and R. L. Vessella Inhibition of Androgen-independent Growth of Prostate Cancer Xenografts by 17{beta}-Estradiol Clin. Cancer Res., April 1, 2002; 8(4): 1003 - 1007. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Ylikoski, M. Karp, K. Pettersson, H. Lilja, and T. Lovgren Simultaneous Quantification of Human Glandular Kallikrein 2 and Prostate-Specific Antigen mRNAs in Peripheral Blood from Prostate Cancer Patients J. Mol. Diagn., August 1, 2001; 3(3): 111 - 122. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Malatesta, F. Mannello, F. Luchetti, F. Marcheggiani, L. Condemi, S. Papa, and G. Gazzanelli Prostate-Specific Antigen Synthesis and Secretion by Human Placenta: A Physiological Kallikrein Source during Pregnancy J. Clin. Endocrinol. Metab., January 1, 2000; 85(1): 317 - 321. [Abstract] [Full Text]
|
|
|
|
|
|
M. D. Ringel, P. L. Balducci-Silano, J. S. Anderson, C. A. Spencer, J. Silverman, Y. H. Sparling, G. L. Francis, K. D. Burman, L. Wartofsky, P. W. Ladenson, et al. Quantitative Reverse Transcription-Polymerase Chain Reaction of Circulating Thyroglobulin Messenger Ribonucleic Acid for Monitoring Patients with Thyroid Carcinoma J. Clin. Endocrinol. Metab., November 1, 1999; 84(11): 4037 - 4042. [Abstract] [Full Text]
|
|
|
|
|
|
A. Ylikoski, M. Sjoroos, A. Lundwall, M. Karp, T. Lovgren, H. Lilja, and A. Iitia Quantitative Reverse Transcription-PCR Assay with an Internal Standard for the Detection of Prostate-specific Antigen mRNA Clin. Chem., September 1, 1999; 45(9): 1397 - 1407. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. L. Cher, J. G. de Oliveira, A. A. Beaman, J. A. Nemeth, M. Hussain, and D. P. Wood Jr. Cellular Proliferation and Prevalence of Micrometastatic Cells in the Bone Marrow of Patients with Clinically Localized Prostate Cancer Clin. Cancer Res., September 1, 1999; 5(9): 2421 - 2425. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Mannello, M. Malatesta, F. Luchetti, S. Papa, S. Battistelli, and G. Gazzanelli Immunoreactivity, Ultrastructural Localization, and Transcript Expression of Prostate-specific Antigen in Human Neuroblastoma Cell Lines Clin. Chem., January 1, 1999; 45(1): 78 - 84. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. D. Ringel, P. W. Ladenson, and M. A. Levine Molecular Diagnosis of Residual and Recurrent Thyroid Cancer by Amplification of Thyroglobulin Messenger Ribonucleic Acid in Peripheral Blood |
