|Home | About | Journals | Submit | Contact Us | Français|
1p/19q (1p and/or 19q) deletions are prognostic factors in oligodendroglial tumors (OT) and predict better survival after both chemotherapy and radiotherapy. While studying 1p/19q status as a potential variable within multivariate prognosis models for OT, we have frequently encountered unknown 1p/19q status within our glioma sample database due to lack of paired blood samples for loss of heterozygosity (LOH) assay and/or failure to perform fluorescence in situ hybridization (FISH). We realized that a 1p and 19q deletion assay that could be reliably performed solely on tumor DNA samples would allow us to fill in these molecular biology data “holes”. We built recombinant DNA with fragments of the selected “marker” genes in 1p (E2F2, NOTCH2), and 19q (PLAUR) and “reference” genes (ERC2, SPOCK1, and SPAG16 ) and used it as quantification standard in real-time PCR to gain absolute ratios of marker/reference gene copy numbers in tumor DNA samples, thus called comparative quantitative PCR (CQ-PCR). Using CQ-PCR, we identified 1p and/ or 19q deletions in majority of pure low-grade oligodenroglioma (OG) tumors (17/21, 81%), a large portion of anaplastic oligodendroglioma (AO) tumors (6/15, 47%), but rarely found in mixed oligoastrcytomas (OA) tumors (1/8, 13%). These data are consistent with results of LOH and FISH assays generally reported for these tumor types. In addition, 15 out 18 samples showed concordant results between FISH and CQ-PCR. We conclude that CQ-PCR is a potential means to gain 1p/19q deletion information, which prognostic and predictive values of CQ-PCR-derived 1p/19q status will be determined in a future study.
The current gold standard for the diagnosis of glioma relies on histological examination of tumor tissue; however molecular analyses have uncovered important genetic abnormalities, some of which provide useful information pertaining to variations in patient’s overall survival and/or response to treatment. Among gliomas, deletion of 1p and 19q are associated with tumors with oligodendroglial components.1,2 1p and/or 19q (1p/19q) deletions have been observed in up to 70% of oligodendrogliomas and 50% of mixed oligoastrocytomas (OA).3,4 In addition to their diagnostic relevance, 1p/19q status is a predictive marker for overall improved outlook for patients with oligodendroglial tumors (OT).3–6 Several studies have suggested that the 1p/19q status (co-deleted or not) predicts response of OT to radiation,7 chemotherapy,8–10 or combined.11,12
Due to the potential importance of 1p/19q status for patients with OT, several molecular methods have been standardized to detect this genetic abnormality in both clinical and basic research settings. The most commonly applied techniques to detect 1p/19q deletion are loss of heterozygosity (LOH),13,14 and fluorescence in situ hybridization (FISH).15,16 LOH requires paired blood DNA from the same patient, which is not routinely collected by most university tissue banks. FISH detection of 1p/19q deletion is frequently employed in clinical laboratories, but is costly and requires highly experienced personnel to ensure accuracy and data standardization.
To date, the best means for incorporating 1p/19q data into clinical practice remains controversial. There is an urgent need for accurate prognostic tools (particularly if the prognostic variables predict therapeutic response), for life-planning purposes for patients and treatment planning purposes for clinicians. However, development of prognosis models for gliomas has been a challenge due to their relatively low incidence and greater variation in survival per patient within a given histology and grade.17 A good statistical percept for building a multivariable prognosis model is that we should have at least 10 events (deaths) for each candidate variable studied.18 Thus when possible all available tumor resource of OT should be utilized. However, as shown in a recent retrospective study of over 1000 WHO grade III OT, 38% of tumor samples were missing 1p or 19q data. As a result, these tumors were excluded from the study.12
To explore an accurate means of filling these 1p 19q “data holes” independent of LOH or FISH via a test solely reliant on tumor sample DNA, we explored real-time PCR technology. We employed a standard-based absolute real-time PCR concept in quantifying gene expression,19 to gain absolute ratios of DNA copy numbers of marker genes in 1p or 19q and reference genes in relatively “stable” areas in chromosomes, here referred as comparative quantitative PCR (CQ-PCR). Over the past decade, the regions of interest on 1p deletion have been mapped to 1p36.3, 1p36.1-p36.2, and 1p34-p35, the region of interest in 19q is located between 19q13.11 and 19q13.4,20,21 and current clinical FISH test probes are associated with 1p36 and 19q13 regions.5,16,22 CAMTA1 and NOTCH2 genes reside in1p36.31-p36.23 and 1p13–11, respectively, with potential tumor suppression functions,23,24 while E2F2 tumor suppressor gene resides in 1p36.
Thus, we selected these three genes to analyze for deletion of 1p in gliomas. We selected PLAUR residing in 19q13 to study DNA copy number variation (CNV) of 19q, for its well-studied function in glioma invasion. According to CGH, there are certain chromosomes that typically remain unaltered during the course of glioma tumorigenesis, including 2q, 3p, 5q, 8q, 12q and 21p.25–27 Hence three genes (ERC2, SPAG16 and SPOCK1) located in chromosomes 3p14.3, 2q34 and 5q31, respectively, were selected as reference genes to determine CNV of the selected marker genes in 1p and 19q. We report a quantitative, sensitive, accurate, and relatively inexpensive assay for determining gene CNV in 1p/19q by real-time CQ-PCR that can be performed using only a few ng tumor DNA sample, independent of the need for LOH or FISH analysis.
Glioma specimens were obtained from patients in accordance with Institutional Review Board approval at each participating institutions (University of California, Irvine, USA and University of Arkansas for Medical Sciences, USA). We included DNA samples from 44 OTs, 9 paired blood lymphocytes collected from patients during surgical resection of their gliomas, and 14 glioblastoma multiformes (GBMs) in this study. Breaking down of 44 OT samples included in analysis are 21 WHO grade II oligodendroglioma (OG), 8 WHO Grade II mixed oligo-astrocytoma (OA), 15 either WHO grade III anaplastic oligodendroglioma (AO) or WHO grade III anaplastic mixed oligo-astrocytoma (AOA). For each case, histological assessment of the tumor tissues were performed by the neuropathologists in the participating institutes. All DNA samples were extracted from frozen glioma specimens using DNeasy kit (QIAGEN), and approximately 2 ng DNA in 10 mM Tris-HCl (pH 7.5) was used to run real time PCR for each gene.
CQ-PCR standard CQ-103 and 10X PCR primer mix for selected marker genes in 1p (CAMTA1, E2F2 and NOTCH2), 19q (PLAUR) and reference genes (ERC2, SPAG16 and SPOCK1); and CQ-PCR standard CQ-108 and 10X PCR primer mix for NLGN4Y and the same set of reference genes are from Ziren Research (Irvine, CA). We run real-time PCR with Roche Light-Cycler real-time PCR instrument 2.0 (Indianapolis, IN) in a capillary tubes with 4 μl of standard or sample DNA (0.5 ng/μl), 1 μl of 10X FAST-START DNA Master SYBR Green I enzyme mix (Roche, Indianapolis, IN), 1 μl of the 10 X primer mix, and 4 μl of water in standard program setting with denature temperature set at 62°C, extension time 12 sec. A melting curve cycle was included for each run to endure no formation of primer dimer after 45 cycles. Gene copy numbers were exported directly from real-time PCR program, which was based on the standard curve (R value >0.99), a regression curve of the “cycle” numbers of four diluted standards when PCR reaching log amplification to their DNA quantifies (102, 103, 104, 105 copies). All genes were quantified at least twice in separate runs, to show assay reproduction with <20% standard deviation (SD), then mean copy number of each gene was taken for analysis, by calculating the ratio of each marker to the reference gene.
Paraffin embedded tissue obtained following surgical resection of glioma from patients were sent to Mayo Clinic, Department of Lab Medicine and Pathology, Rochester, Minnesota, USA for performing FISH with test probes (1p36 and 19q13) and control probes (1q24 and 19p13). FISH signals for each gene-specific FISH probes were assessed under fluorescence microscope and the ratio of the probes for 1p/1q and 19p/19q were enumerated in 100 non-overlapping nuclei. A normal ratio is considered 1.0 and any ratio <0.80 is considered deletion of the region of interest.
The principle of CQ-PCR as a modification of single-copy-gene standard-based real time PCR is via using a multi-gene-containing recombinant DNA standard that allows determination of absolute ratios of marker to reference gene copy numbers in sample DNA. A ratio of 1:1 between selected marker and reference genes in autosomal chromosomes is expected in normal cells while changes in this ratio in tumor DNA would suggest CNV, either deletion or amplification, in the studied gene of interest.
Considering the inherent genome instability of cancer cells, we analyzed the stability of three reference genes in tumor samples and found amplification of SPAG16 in some OT. To mitigate this, we took the average of two ratios of ERC2 and SPOCK1 for most tumors. For other samples, the two reference gene ratios showing the most concordance were used to take a mean and SD.
With consideration of 10%–20% variation inherited with real-time PCR, the mean values of the marker and reference ratio was taken for determination of deletion (<0.8) or amplification (>1.2), Shown in Table 1, there was a gain at the 1p marker gene CAMTA1 (1p36.31-23) in both GBM and OT, which were not found in other two 1p marker genes E2F2 (1p36) and NOTCH2 (1p13-p11). Thus average of these two 1p marker genes ratio to reference gene were taken to determine 1p deletion status (value <0.80 is considered 1p deleted). Based on the results from 67 samples, the mean (medium) of the SDs from different reference ratios for the four marker genes is 10%–15% (7%–9%), which supports our setting of cut-off-line of 0.8 or 1.2 in determining gene CNV from results of CQ-PCR.
Similarly, the ratio of DNA copy number between a gene in Y-chromosome and a gene in autosomal chromosome is expected to be 0.5 in a normal male diploid genome while zero in female DNA. As shown in Figure 1. CQ-PCR determination of absolute ratio of two gene copy numbers have given rise to a ratio value of 0.4–0.5 in normal cells of male and zero in female between genes in Y-(NLGN4Y) and autosomal chromosomes, while 1.0 between two autosomal genes (ERC2, PGCP, SPAG16 and SPOCK1). To control human error in managing the tissue bank, we first carried out CQ-PCR of NLGN4Y and the autosomal genes in 61 tumor and 9 paired blood DNA samples, and analyzed for concordance with gender information provided by each participating institute’s tissue bank. Out of total 70 DNA samples we identified 3 tumor samples with conflicting information between gender-associated gene in DNA and gender information of the patients, which were excluded from this study.
Blood sample is generally considered to be normal, in terms of mutations and allelic loss or amplification in the tumor sample. It is the reference used by loss of heterozygosity assay (LOH) for detecting loss of a microsatellite allele. In our CQ-PCR, we selected single copy protein-coding gene for both marker and reference genes, detected the copy numbers of each, and compared relative difference in quantity. If there is deletion or amplification of a gene that is associated with tumor, it normally should not occur in normal lymphocyte. Our CQ-PCR detected deletions of 1p (E2F2, NOTCH2) and 19q (PLAUR) markers in 5/9 OT but none in the paired blood DNA samples; results for 6 tumor DNA samples with their paired blood DNA 1p and 19q were shown in Figure 2. CAMTA1 was deleted in two and amplified in other two of the the six OT, but not in the paired blood DNA samples.
Based on average of E2F2 and NOTCH2 for 1p and PLAUR for 19q, as shown in Tables 1 and and2,2, our CQ-PCR revealed 1p/19q deletion for 57% OTs overall (25/44). Deletion of the 1p marker genes were not seen in the 14 GBM DNA samples, while deletion of the 19q marker gene was found in 8 out 14 (57%) GBM samples but not found in the eight OA samples. Thus, we did not find co-deletions of 1p and 19q in our GBM or OA samples. Deletion of either 1p, or 19q marker genes were mainly found in OG (48% and 76%, respectively). Co-deletions of 1p and 19q were also mainly found in OG (9/21, 43%), compared with AO & AOA combined (only 1 of 15, 7%).
The attempt of this study is to develop a PCR-based detection of 1p19q deletion for use in predicting patient’s outcome. While FISH-based 1p19q deletion data have being shown with values in prognosis and prediction of radiation and chemotherapy, we therefore compared our CQ-PCR-derived data with that of FISH. Shown in Table 1, out of 18 DNA samples which had 1p19q deletion status available by both CQ-PCR and FISH techniques, the concordance rate between the two techniques was 83% (15/18) and 78% (14/18) for 1p and 19q deletions, respectively.
Our CQ-PCR detected 1p and/or 19q deletions in majority of classical oligodendroglioma samples (81% of OG and 47% of AO), but in minority of OA (13%). These results data consistent with those generally reported for LOH and FISH assays studying these tumor types.3,4,23 In addition, 15 out 18 samples showed concordant results between FISH and CQ-PCR. Considering the small tumor sample size and limited marker loci used in our study, such a high concordance is remarkable, and suggests that our CQ-PCR has the potential to be used to define 1p/19q deletion status independent of FISH and LOH analysis. CQ-PCR does not require paired blood samples, strong reliance on technical considerations such as technician consistency and experience, and is much less costly than either LOH or FISH. Thus we could use this technology to effectively screen all available tumor DNA samples to determine 1p/19q deletion status which will aid in building multivariate prognosis models for OT. The prognostic and predictive values of CQ-PCR-derived 1p/19q status will be determined based on a larger sample size with a large enough event number.
Our CQ-PCR finding of no 1p deletions in GBMs is consistent with results reported with LOH and FISH. However, we did detect PLAUR deletion in19q in 57% of the GBM samples, which has not been previously recognized. In contrast to GBM samples, none of the 8 OA samples had deletion of PLAUR. The identification of gain at 1p marker genes CAMTA1, E2F2 and NOTCH2 and 19 q marker gene PLAUR was not the focus of this study; however, in many OT and GBM samples, gains were detected by CQ-PCR that were not evident by FISH. Their biological implications and prognostic/predictive values on survival and treatment response are yet to be determined.
There have being prior explorations on obtaining 1p/19q status based solely on tumor DNA. High throughput genomic DNA microarray-based technology known as comparative genomic hybridization is able to detect 1p/19q deletions in gliomas, in addition to providing information on DNA CNV in the entire genome.28,29 It requires well trained personnel to not only perform the assay, but also process the data; thus incurring high cost in detecting a few known marker loci. Other PCR-based assays have also emerged including quantitative microsatellite analysis (QUMA) 30,31 and multiplex ligation-dependent probe amplification (MLPA).32 QUMA uses PCR amplification of microsatellite loci that contain (CA)n repeats to determine CNV of genes of interest, and has also been validated with FISH technology on detection of 1p and 19q loss in oligodendroglioma.31 However use of primers to amplify microsatellite DNA in quantitative PCR can be noisy, as multiple lengths of PCR product can be produced from cells even with homozygous loci, and amplification efficiency of different microsatellite markers can be different, thus the comparability between different platforms would be challenged. MLPA-based detection of 1p/19q has the advantage of detecting copy number changes of up to 45 loci in one relatively simple PCR based assay but it is semi quantitative and has yet to be widely adopted.32
Our real-time CQ-PCR utilizes multi-gene recombinant standard DNA to determine the absolute ratio of DNA copy numbers between the marker genes and multiple reference genes in tumor specimens. Both marker and reference genes were selected for their locations of interest and single-copy in genome with PCR primers designed to avoid amplification of corresponding pseudogenes in genome. It has advantages offered by QUMA in that all loci are informative, paired normal tissue from the same patient is not required, and even gain can be distinguished from loss. In comparison to QUMA, which requires accurate input of tumor DNA quantity, result from CQ-PCR is an absolute ratio of two gene’s DNA quantity. Thus the result is independent of DNA input quantity and the accuracy is ensured by the specificity of PCR primers.
Our results showed that CQ-PCR is a highly sensitive test; 100 copy numbers of a specific gene were able to be robustly quantified in highly diluted DNA samples (0.5 ng/μL). In addition to determine loss (here 1p/19q), gain of gene copy numbers, such as EGFR was also be detected by CQ-PCR (our unpublished data).
Inherited with PCR-based molecular assay, this technique has the disadvantages that it will not detect events without causing an alteration of DNA copy number, such as chromosomal translations, and it relies on the stability of reference genes. The former concern is not an issue for detecting deletion of 1p/19q, while the later has been considered by selecting multiple reference genes residing in relatively stable areas of genome in specific cancer types, which information is based on whole genome comparative analyses. It still needs to be empirically validated based on a large number of tumor samples. In this study, we have tested four candidate reference genes in total 58 gliomas, with ERC2, SPOCK1, and/or SPAG16 giving similar results for the marker gene on CNV, while not when normalized with PGCP, a reference gene selected based on stable area of chromosome (data not shown). In this study, we have taken average ratios to ERC2, SPOCK1, and/or SPAG16 to determine deletion status of 1p/19q marker genes. The average (medium) of the SDs from different reference ratios for the four marker genes is 10%–15% (7%–9%), which verified our proper use of the reference genes.
In conclusion, CQ-PCR technology reported here could be used to effectively screen all available tumor DNA samples to get 1p/19q deletion status. Our prior studies of glioma prognosis have revealed a prognostic model for OT with high predictive accuracy that explained 63% survival variation with significant likelihood ratio P-value = 0.0076.33 It is yet to be determined if the CQ-PCR-detected 1p/19q deletion has prognostic value, and furthermore if it improves the multivariable OT models we have established with combined clinical and gene expression variables.
This work was supported by UC Irvine CORCL and Chao Family Comprehensive Cancer Center Seed Grant.
Author(s) have provided signed confirmations to the publisher of their compliance with all applicable legal and ethical obligations in respect to declaration of conflicts of interest, funding, authorship and contributorship, and compliance with ethical requirements in respect to treatment of human and animal test subjects. If this article contains identifiable human subject(s) author(s) were required to supply signed patient consent prior to publication. Author(s) have confirmed that the published article is unique and not under consideration nor published by any other publication and that they have consent to reproduce any copyrighted material. The peer reviewers declared no conflicts of interest.