| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Article |
Department of General Surgery, Duke University Medical Center, Durham, North Carolina 27710
Correspondence: Address correspondence and reprint requests to: Scott K. Pruitt, MD, PhD, FACS; E-mail: scott.pruitt{at}duke.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods: The prostate stem cell antigen (PSCA) gene was identified as a marker highly expressed in pancreatic adenocarcinoma and not normal pancreas. RNA from pancreatic and nonpancreatic cancer cell lines as well as tissue and blood from pancreatic cancer and control patients was reverse-transcribed and PSCA quantified by qPCR.
Results: Individual operator experience affects the results of qPCR, with significantly different copy numbers at experiment numbers 5, 15, and 40. Five of six pancreatic cell lines had PSCA/actin ratios 10-fold greater than nonpancreatic cancer lines. Mean PSCA expression in pancreatic tumor tissue was significantly higher (P < 0.05, Student’s t-test) than in the tissue of benign pancreatic processes. The close correlation of PSCA/actin copy number with number of tumor cells in the blood was demonstrated by regression analysis (r = 0.768, P = 0.0001). PSCA copy number was significantly higher in the blood of patients with metastatic pancreatic cancer than in that of normal patients (P < 0.05, Student’s t-test).
Conclusions: Such trends suggest that PSCA may prove to be a valuable pancreatic cancer tumor marker. More generally, the technique of qPCR is shown to provide a sensitive method of evaluating markers in cancer patients.
Key Words: Quantitative PCR • Pancreatic adenocarcinoma • Tumor marker • Prostate stem cell antigen
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Identifying a specific and sensitive pancreatic tumor marker could prove valuable in the diagnosis and management of pancreatic cancer patients. In this group of patients, goals of a tumor marker would include determination of diagnosis at an early stage to allow for definitive therapy as well as help to confirm diagnosis when a tissue specimen is not available. In addition, the detection of subclinical tumor in the blood may prove important in not only identifying groups at high risk for recurrence after definitive treatment but also diagnosing recurrent disease earlier than currently utilized imaging studies. Finally, monitoring response to treatment, both in the adjuvant and metastatic disease settings, could be useful in determining response to therapy.
The molecular technique of reverse-transcription polymerase chain reaction (RT-PCR) has emerged as a promising evaluator of tumor markers because of its ability to allow cancer detection at an earlier stage when tumor burden is small and potentially more treatable.4 Conventional RT-PCR offers the advantages of greater sensitivity and wider variety of targets that can be investigated compared to older techniques such as immunohistochemistry and has been used in the past to evaluate potential pancreatic tumor markers.5 However, conventional PCR has several major drawbacks, including its ability to generate only qualitative or, at best, semiquantitative data. In addition, there are contamination risks associated with the multiple steps and widely reported variability of results, leading to this technique being unreliable as a clinical tool. More recently, the use of real-time quantitative PCR (qPCR) has emerged as a potential tool to identify subclinical tumor burden. This technique has the benefits of determining precise levels of target nucleic acids, reliably measuring in real time as each cycle occurs, and fewer processing steps. While the potential for real-time qPCR with its conceptual simplicity,6 practical ease,7 and high-throughput ability8 have made it quite popular of late, it comes with its own set of potential pitfalls, including variability9 and reproducibility10 issues.
After selecting a new candidate tumor marker for pancreatic adenocarcinoma utilizing the technique of serial analysis of gene expression, the goal of this study was to determine reproducibility and variability of real-time qPCR. Specifically, using a pancreatic tumor marker as the target of interest, we attempted to characterize the level of expression of this marker in pancreatic cancer cell lines as well as cancerous and benign pancreatic tissue and to assess the ability of qPCR to detect pancreatic tumor cells in peripheral blood.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Pancreatic tissue was evaluated in 13 patients undergoing surgery for pancreatic carcinoma and in six patients undergoing surgery for benign pancreatic conditions including benign cysts (n = 3) and chronic pancreatitis (n = 3). Fresh tumor specimens were obtained by core needle biopsy at the time of surgery, and diagnosis was confirmed by histopathology. All specimens were obtained and reviewed by DUMC surgeons and pathologists under a protocol approved by the DUMC Institutional Review Board. Upon biopsy, the tissue was immediately placed in RNAlater (Ambion, Austin, TX) and stored at 4°C. Within 2 weeks of biopsy, a 30-mg sample of tissue was snap-frozen on dry ice and 1,000 µL of RLT buffer (Qiagen RNAeasy kit) was added, followed by homogenization. One thousand microliters of 70% ethanol were then added to the solution, and the total RNA was extracted using the RNAeasy kit.
Blood Collection
Twenty milliliters of blood were drawn into vacutainers containing ethylenediaminetetraacetic acid (EDTA) after at least a 2-mL waste to avoid contamination with skin flora. Peripheral blood mononuclear cells (PBMCs) were isolated using Purescript RBS lysis solution (Gentra Systems, Minneapolis, MN), centrifugation, and multiple washes with PBS. Total RNA extraction of PBMCs was then carried out with the RNAeasy kit.
Blood was drawn from 47 pancreatic adenocarcinoma patients at various stages of therapy, specifically after diagnosis of resectable disease but prior to any treatment (n = 10), post-neoadjuvant therapy with 5-fluorouracil (5-FU) and external radiation and prior to surgery (n = 10), postsurgery (pancreatico-duodenectomy) and prior to adjuvant therapy (n = 11), and postsurgery and post-adjuvant therapy with 5-FU and external radiation (n = 7), as well as from patients with metastatic disease at the time of diagnosis (n = 9). Blood from patients with benign pancreatic illnesses such as benign cysts and pancreatitis (n = 4) and healthy patients with no diagnosed pancreatic illness (n = 4) was also drawn. All blood draws were under a protocol approved by the DUMC Institutional Review Board.
RNA Extraction and cDNA Synthesis by Reverse Transcription
Total cellular RNA from cell lines, blood, and tissues was extracted and isolated using the Qiagen RNAeasy kit. All RNA extractions were carried out in a designated hood with RNase-free labware. Purified RNA was quantified by spectrophotometry (Beckmann Instruments, Fullerton, CA). Tissue processing, RNA extraction, reverse transcription (RT), and real-time PCR assay setup were performed in separate areas and rooms as well as at separate times to prevent crossover contamination. RT was performed using Moloney murine leukemia virus Superscript II RNase H– Reverse Transcriptase (Invitrogen, Carlsbad, CA). Total RNA (2 µg) were used for each reaction for all samples in the study. All RT reactions were carried out with oligo(dT)12–18 priming (Invitrogen) in order to target transcription of polyadenylated mRNA. The resulting cDNA was then frozen at –20°C until use in real-time PCR.
Identification of Pancreatic Tumor Markers and Creation of Primers
Serial analysis of gene expression (SAGE) is a technique that provides a comprehensive and quantitative profile of cellular gene expression.11,12 In this technique, cellular mRNA transcripts are converted to cDNA and cleaved at specific sites by restriction enzymes into 10–14 bp tags. Each tag should uniquely identify a specific gene transcript,11 and the abundance of each tag provides a quantitative measure of the transcript level in the mRNA sample analyzed. This ability allows the expression levels of specific transcripts to be compared between two samples.12 SAGE may be performed by accessing the xProfiler program, available online through the National Center for Biotechnology Information (NCBI) at the National Institutes of Health website (http:// www.ncbi.nlm.nih.gov/SAGE). We queried the program to compare gene expression between a pancreatic adenocarcinoma group and a nonneoplastic tissue group. The pancreatic adenocarcinoma group consisted of 202,237 tags made up of gene transcripts from pancreatic cancer cell lines (SAGE libraries CAPAN1, CAPAN2, Hs 766T, Panc-1) and primary pancreatic carcinomas (Panc-91-16113, 96-6252). The nonneoplastic group involved 537,681 tags from normal pancreas epithelium (H126, HX) as well as normal colonic (NC1, NC2), prostatic (Chen Normal Pr, normal prostate), ovarian (HOSE 4, IOSE 29-11), mammary (mammary epithelium, Br N), and microvascular epithelial (Duke HMVEC, HMVEC-VEGF) cell lines. We ordered the analysis to display the 100 SAGE tags that showed a >10-fold difference in expression levels between these two groups, of which prostate stem cell antigen (PSCA) was chosen for further analysis.
The primers and probe for PSCA were designed using the applications-based software Primer Express (Applied Biosystems, Foster City, CA). The sense primer was 5'-GTGGATGACTCACAGGACTACTACGT-3' and the antisense primer was 5'-CGCTGGCGTTGCACAAG-3'. The probe was 6-carboxyfluorescein (FAM)-5'-AAGAAGAACATCACGTGCTGTGACACCG-3'-6-carboxytetra methylrhodamine (TAMRA). These primers and probes were then analyzed by the NCBI BLAST program. The PSCA design was based on the sequence from GenBank (accession NM_005672). Primers and probes were synthesized by Applied Biosystems. Actin primer and probes were purchased from Applied Biosystems as well.
Real-Time qPCR
Background
Absolute quantitation of PSCA mRNA was achieved using the ABI Prism 7900 Sequence Detection system (Applied Biosystems). In real-time quantitation technology, the 5' exonuclease activity of the Taq polymerase cleaves and releases the hybridization probe that is labeled with a 5' fluorogenic dye (FAM in this case). In its native form, the 5' fluor is quenched by a second dye on the 3' end of the probe (TAMRA in this case) and only upon cleavage is it able to fluoresce freely.13 This probe is specific for a target sequence between the forward and reverse primers and thus generates a fluorescent signal that is specific and directly proportional to the amount of PCR product synthesized. Quantitative RT-PCR allows the measurement of PCR product accumulation during the exponential phase of the reaction as it occurs; the more abundant the initial quantity of a target, the earlier the PCR will be detected by means of a fluorescent signal. PCRs are characterized by the time point during cycling when amplification of the product is first detected, not by the amount of product accumulated after a certain number of cycles.14 In this technology, the target quantity is measured by identifying the cycle threshold (Ct), which is the PCR cycle number at which the reporter fluoresces at or above a preset threshold, in this case at an amount 10 times greater than background.
Creation of Standard Curve
In absolute qPCR, absolute quantitation of transcription allows the precise determination of copy number per cell.15 It requires the construction of an absolute standard curve of known amounts of each gene of interest to quantify the unknowns of interest. In this study, cDNA standards were constructed consisting of plasmid standards for PSCA and actin. Total RNA was extracted from the Hs 766T cell line, and cDNA fragments were generated by RT-PCR using the same primers for PSCA and actin as described above. The amplicons underwent ethidium bromide gel electrophoresis and were analyzed for appropriate sequence length, and the DNA was extracted from the gel and purified with the QUIquick gel extraction kit (Qiagen, Valencia, CA). Each of these amplicons was cloned into a separate pCR2.1-TOPO® (Invitrogen). Ligated fragments were transformed into OneShot® Chemically Competent Escherichia coli cells (Invitrogen), and plasmid DNA was purified using a modified alkaline lysis procedure followed by binding of the plasmid DNA to an anion-exchange resin (Plasmid MaxiKit; Qiagen, Valencia, CA). cDNA plasmid concentrations were measured by optical density spectrophotometry, and the corresponding copy number was calculated. Serial dilutions from the resulting clones were used as standard curves, each containing a known amount of input copy number. One master set of dilutions was made to be used in all PCR runs. These were aliquoted and frozen at –80°C until use.
Real-time PCR Procedure
In each MicroAmp optical tube of a 96-well plate (Applied Biosystems), 25 µL of a final PCR mixture were placed. This mixture included 1 µL of cDNA template, 3.5 mM MgCl2; 200 µM each of dATP, dCTP, and dGTP; 400 µM dUTP, 0.05 U/µL AmpliTaq Gold DNA polymerase, and 0.01 U/µL AmpErase UNG to prevent PCR product carryover. The mixture also contained the passive reference dye ROX, which provides an internal reference to which the reporter dye signal can be normalized, compensating for the fluorescence between wells and between experiments caused by errors in pipetting or instrument variability. Also included in the PSCA standard and sample wells was 600 nM each of forward and reverse PSCA primer as well as 300 nM PSCA probe. In the actin standard and sample wells, the mixture contained 300 nM each of forward and reverse actin primer and 200 nM actin probe. All primer and probe concentrations were determined by optimization studies. The 96-well plate was then covered with an Applied Biosystems Prism optical adhesive cover. Every PCR run contained triplicates of a seven-point PSCA standard and a five-point actin standard to generate standard curves for PSCA and actin, as well as a triplicate of no template controls. Samples were run in triplicate.
Using the Applied Biosystems Prism 7900 Sequence Detector, the PCR began with a 50°C, 2-minute cycle to optimize UNG activity, followed by a 95°C, 10-minute cycle to activate AmpliTag Gold Polymerase and deactivate UNG. Then, 40 cycles at 94°C for 15 seconds and 60°C for 60 seconds were performed. No post-PCR handling occurred.
Calculation
For each unknown test sample, amounts of PSCA and actin were determined from the respective standard curve. Averaging the triplicates and then dividing the PSCA level by the actin level resulted in a normalized PSCA value
Operator Experience Variability and Reproducibility Studies Using Real-Time PCR
To assess operator variability, the same operator used identical micropipettes, reagents, PSCA primers/probe set, and RNA template (Colo 357) to perform real-time qPCR for the fifth, fifteenth, and fortieth time performing the procedure. The variability of the triplicates within runs and the average quantities between runs were examined. The reproducibility of all steps of the real-time RT-PCR technique combined was evaluated by taking three separate collections of the same number of Colo 357 cells and individually extracting RNA, reverse transcribing, and performing real-time qPCR. These three experiments were performed by the operator after the 40 qPCR experiments described above.
Tumor and Blood Specimen Evaluation
Patient tissue cDNA real-time PCR was performed with the operator blinded to the samples. To examine the real-time PCR technique’s ability to measure circulating tumor in the blood, a 1-mL aliquot of control patient’s blood was spiked with 10 tumor cells to 1 x 106 tumor cells with subsequent RNA extraction, reverse transcription, and real-time qPCR. Patient blood cDNA real-time PCR was then performed with the operator blinded to the samples.
Statistical Analysis
Regression analysis was used to determine correlations between logarithmic scaled standard copy amounts and the linearly scaled standard Ct values. The coefficient of variance was used to describe intraassay and interassay precision for the actin and PSCA standard curves. The Kruskal-Wallis test was used to assess operator variability and one-way analysis of variance (ANOVA), to evaluate reproducibility. Regression analysis was used to evaluate correlation between tumor cell and copy number in the spiking experiments. Student’s t-test was used to compare PSCA copy numbers in the blood and tumors of patients with pancreatic cancer with those of the control groups.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Standard Curve
Using PSCA as our target DNA, we then set out to establish a standard curve for this marker. The absolute copy amount of an unknown DNA sample can be calculated from the regression curve of the logarithmic scaled standard copy amounts vs. the linearly scaled standard Ct values, an example of which is seen in Figure 1
. R2 for PSCA was 0.977 (± standard deviation [SD] 0.02) and that for ß-actin was 0.980 (±0.01). Intraassay coefficients of variance (CV) were 2.0% (PSCA) and 6% (ß-actin). Interassay CV were 9.3% (PSCA) and 10% (ß-actin) as calculated by variation from the mean of six separate measurements of the same DNA.
|
with SDs within triplicates of each group of 0.97, 0.67, and 0.12 for experiments 5, 15, and 40, respectively. These threshold values translated into significantly different copy numbers of 2.67 x 103, 2.56 x 105, and 3.06 x 105 with a P value of 0.06 using the Kruskal-Wallis test.
|
As seen in Figure 3, the reproducibility results showed no statistically significant difference in copy number between the three groups of 7 x 106 Colo 357 cells prepared separately when examined by one-way ANOVA.
|
).
|
).
|
, has an R2 = 0.7879 with y = 9E – 07e0.4943x.
|
. In comparing individual groups by Student’s t-test, the metastatic group’s gene expression is significantly higher, with P < 0.05, than the normal group as well as the postneoadjuvant/presurgical and postsurgical/ preadjuvant groups. These observations suggest that one potential application of a tumor marker using this technique would be to evaluate subclinical disease burden.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
When using real-time RT-PCR, one must remember that the single most likely source of data variation is variability introduced by the person carrying out the experiment.6 The variability of results from identical samples assayed in different laboratories emphasizes this problem.9 This variability is emphasized in our study when the only factor manipulated was operator experience (five experiments worth of experience compared with 15 and 40). The SD within triplicates decreased with experience, as would be expected. Alarming is the significantly different corresponding PSCA copy quantity of the same cDNA when only handler experience was modified. These data support the idea that, all else equal, this technique is operator experience-dependent. One criticism of this experiment setup is that the cDNA template was stored at –80°C between experiments and could have degraded over the 4 months between runs 5 and 40. While this is true, one would think the copy number would have decreased rather than increased, as was the case. Once the technique had been appropriately mastered, the amount of variability of the standard curves both intraassay and interassay, with CV of 2% and 9%, respectively, was monitored and was within this low range.
When using absolute qPCR where a standard curve with known copies of a gene is used to extrapolate the unknowns, it is not necessary to include a control gene.13 However, we chose to use such a control to help account for differences in RNA extraction and cDNA synthesis. The RT step has been proposed as a large source of variability in qPCR experiments because of the sensitivity of RT enzymes to salts, aliphatic alcohols, and phenols often used during the RNA isolation step.26 In addition, RT-PCR-specific errors are compounded by any variation in the amount of starting material, especially when the samples are obtained from different individuals.15 An internal reference is an accepted method for minimizing these errors and correcting for sample-to-sample variability. Unlike in relative qPCR where the quantitation is done relative to the control gene under the assumption that the chosen control gene does not vary in copy number or expression among samples, we performed absolute quantitation with the control gene as well as with the gene of interest. We chose ß-actin as our control gene because it is expressed at moderately abundant levels in most cell types and has been advocated as a quantitative reference.27
We thrice independently extracted and prepared RNA from separate aliquots of the same number of cells and performed RT and subsequent real-time PCR, with the results showing no significant difference in copy number between the separate assays. From these data, we conclude that the reproducibility of this technology as a whole, despite the pitfalls present at each step, is acceptable in our hands. Individual labs that perform real-time qPCR should have mechanisms in place for testing operator variability and assay reproducibility before proceeding with this technique.
The increased expression of PSCA in pancreatic cancer cell lines compared to other neoplastic cell lines lends support to the idea that this marker, when expressed above a certain threshold, could be specific to this disease. Only through the ability of real-time PCR to give us quantitative data were we able to make this distinction. In comparing pancreatic adenocarcinoma tumor gene copy number to that of benign pancreatic tissues, again there appeared to be a threshold expression level above which only adenocarcinomas resided. Though the numbers studied are small, the trend supports the idea that within the pancreas elevated PSCA expression may be specific to carcinoma and, thus, an appropriate pancreatic tumor cancer tumor marker using this technique.
In order to have clinical utility as a marker of disease diagnosis and progression, expression of the gene of interest must be easily monitored. We demonstrated the ability of real-time PCR to detect pancreatic tumor cells in the blood by determining PSCA expression in blood spiked with different numbers of cancer cells. The correlation between tumor amount and PSCA/actin copy number was meaningful, as seen in Figure 6
with R2 of 0.7879. PSCA/actin expression in the blood of patients with metastatic disease was significantly higher than that of normal patients as well as those who had undergone neoadjuvant therapy and were awaiting surgery and those who had undergone surgery and were scheduled to have further adjuvant treatment. These data lend support to the use of PSCA expression in the blood as a marker of subclinical disease burden. By quantifying tumor cells in the blood during the course of treatment and follow-up, substantially more information on disease status will be gained.
Because PSCA mRNA does not appear to be expressed in the peripheral blood or target organ, its detection is indicative of the presence of tumor cells. The ultimate ability to assess PSCA as a marker of pancreatic cancer will be whether its quantity in the blood has prognostic importance in disease progression and survival. We plan to follow these patients’ clinical courses and make such a determination. Further studies using serial blood samples from individual patients going through treatment will be required before definitive judgments about this marker and qPCR can be made.
Our original observation suggests that real-time qPCR allows assessment of a tumor marker in a manner not possible using other techniques. This technique has a promising role in this expanding field, as long as its limitations are taken into account.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication June 18, 2003. Accepted for publication April 12, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
