METHODS

Blood samples

Venous peripheral blood samples were taken from patients during surgery for an astrocytic glioma of the brain. Blood was collected in citrate in aliquots of 3 ml and frozen at −80°C until required. Total cellular RNA was extracted from whole blood using the QIAmp RNA blood mini kit (Qiagen, Hilden, Germany). RNA was treated with DNase I (Qiagen) during extraction for 15 minutes at room temperature to remove any contaminating DNA before RT-PCR. Total RNA was eluted from columns with 30 μl of RNase free water. These procedures were performed according to the manufacturer’s recommendations. The RNA concentration and purity were determined photometrically at 260 and 280 nm, respectively. The average concentration of the extracted total RNA was 20 ng/μl. A 2 μl aliquot of the eluate was taken for each RT-PCR reaction. Total RNA isolated from human glioblastoma tissue, which was immunhistochemically shown to express GFAP, served as a positive control. The histological diagnoses of gliomas were made according to the most recent World Health Organisation classification, and included five diffuse (fibrillary) astrocytomas (grade II), five anaplastic astrocytomas (grade III), and 10 glioblastomas (grade IV). More than 30% of tumour cells showed immunohistochemical expression of GFAP in all tumours. Patients’ informed consent and institutional ethical approval were obtained.

RT-PCR assay

The one step real time RT-PCR assay was performed using fluorescent hybridisation probes and fluorescence resonance energy transfer for the detection of PCR amplification products.⁶ Briefly, primers and probes were designed to amplify a 159 bp (GFAP) and a 108 bp (β2 microglobulin; B2M) specific PCR product, where the B2M microglobulin housekeeping gene served as a positive control. Amplification was performed using the LightCycler hybridisation master kit on a LightCycler (Roche, Grenzach, Germany). Reverse transcription was performed at 61°C for 30 minutes, followed by an initial denaturation step at 95°C for 30 seconds and by 45 cycles of PCR. Each PCR cycle consisted of a one second denaturation step at 95°C, followed by a 15 second annealing step at 55°C, and an elongation step at 72°C for 13 seconds. Temperature transition rates were set to 20°C/second (denaturation and annealing) and 2°C/second (elongation). The following primers and probes were used.

• For GFAP: 5′-CGATCAACTCACCGCCAACA-3′ (forward primer), 5′-GTGGCTTCATCTGCTTCCTGTC-3′ (reverse primer), 5’-AGAAGCTCCAGGATGAAACCAACC-Fl (donator probe), and LC640-GCTGGAAGCCGAGAACAACCTG (acceptor probe).

• For B2M: 5′-AAGATGAGTATGCCTGCCGTG-3′ (forward primer), 5′-ATGCGGCATCTTCAAACCTC-3′ (reverse primer), 5′-GCTGCTTACATGTCTCGATCCCAC-Fl (donator probe), and LC705-AACTATCTTGGGCTGTGAGAAAGTCA (acceptor probe).

The assay was titrated using human U373 glioma cells and shown to detect an RNA amount corresponding to one GFAP positive cell among 10⁶ GFAP negative cells.

RESULTS AND DISCUSSION

Real time RT-PCR revealed the absence of GFAP transcripts in the blood from all 20 patients with glioma during 45 cycles (fig 1A, B). The positive control (glioblastoma tissue) exhibited strong amplification, confirming that the assay conditions were correct (fig 1A,B). Agarose gels of the amplification product of the positive control revealed a single band at 159 bp, excluding additional non-specific amplification (not shown). Expression of B2M mRNA in all samples indicated intact mRNA and similar amounts of mRNA among the specimens (fig 1C,D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Fluorescence intensities, reflecting amounts of (A, B) glial fibrillary acidic protein (GFAP) mRNA and (B2M) β2 microglobulin mRNA (C, D), are shown as a function of the cycle number. Each line represents one sample. The single specimens in (A) and (B) showing strong amplification represent the positive control (glioblastoma tissue). Amplification of B2M mRNA in all specimens confirms the presence of similar amounts of intact mRNA (C, D). (A, C) patients with astrocytoma; (B, D) patients with glioblastoma.

“Our data suggest that even surgical resection does not result in the washing of tumour cells into the bloodstream”

The absence of detectable amounts of GFAP mRNA in the blood of patients with astrocytoma and glioblastoma, reflecting the virtual absence of glioma cells, supports the notion that intravasation represents a major obstacle for glioma cell metastasis. Because we obtained blood from patients during surgery and previous evidence indicates that systemic metastasis primarily occurs after neurosurgical intervention,² our data suggest that even surgical resection does not result in the washing of tumour cells into the bloodstream, as shown with the sensitive RT-PCR assay used here. However, we cannot entirely exclude the dissemination of GFAP negative tumour cells. Thus, GFAP negative tumour cells, encountered in most astrocytic gliomas in variable numbers, and possibly reflecting cellular anaplasia, could have a particular propensity to metastasise. Furthermore, because GFAP expression is regulated by extracellular signals and increased by extracellular matrix components,⁷ it is possible that GFAP is downregulated in the blood and upregulated again after extravasation, but experimental evidence supporting this speculation is absent. Most systemic glioblastoma metastases continue to express GFAP as determined by immunohistochemistry,^{3, 8} although the loss of GFAP expression has also been described in a transplantation case.⁴ Clearly, the assay described here is not useful for detecting mRNA derived from tumours with a low degree of GFAP expression, including oligodendrogliomas and primitive neuroectodermal tumours.

There is some evidence that gliomas that metastasise are somewhat different from the more typical non-metastatic gliomas, because metastatic glioblastoma occurs in relatively young patients, with a mean age of 38 years, and TP53 mutations tend to be overrepresented in metastatic primary glioblastomas.⁸ Interestingly, TP53 mutations were discordant between cerebral and metastatic glioblastomas in two of four cases analysed, suggesting that distinct mutations occurring in tumour subclones may underlie the propensity to metastasise.⁸ Our data also support the hypothesis that systemic metastasis of astrocytic gliomas, even at subclinical levels, is restricted to distinct genetic subsets of gliomas and represents a very rare event. It will be crucial to determine the genetic or phenotypic features predisposing to metastasis so that gliomas with high metastatic potential can be identified prospectively, which should benefit both patients with glioma and transplant recipients.