Association between laminin-8 and glial tumor grade, recurrence, and patient survival

Authors


  • The antibody C4 to the laminin β2 chain produced by Dr. Joshua Sanes was obtained from the Developmental Studies Hybridoma Bank, Department of Biology, University of Iowa (Iowa City, IA), under contract N01-HD-2-3144 from the National Institute of Child Health and Human Development (NICHD).

Abstract

BACKGROUND

The authors previously sought to identify novel markers of glioma invasion and recurrence. Their research demonstrated that brain gliomas overexpressed a subset of vascular basement components, laminins, that contained the α4 chain. One of these laminins, laminin-8, was found to be present in highly invasive and malignant glioblastoma multiforme (GBM) (Grade 4 astrocytoma); its expression was associated with a decreased time to tumor recurrence, and it was found in vitro to promote invasion of GBM cell lines.

METHODS

In the current study, the authors studied glial tumors of different grades in an attempt to correlate laminin-8 expression with tumor recurrence and patient survival. Immunohistochemistry and Western blot analysis were used to detect laminin isoforms of interest.

RESULTS

Using immunohistochemistry and Western blot analysis, the authors confirmed high levels of laminin-8 expression in approximately 75% of the GBM cases examined and in their adjacent tissues, whereas astrocytomas of lower grades expressed for the most part a different isoform, laminin-9, which also was found in low amounts in normal brain tissue and benign meningiomas. Overexpression of laminin-8 in GBM was found to be associated with a statistically significant shorter time to tumor recurrence (P < 0.0002) and a decreased patient survival time (P < 0.015).

CONCLUSIONS

The data suggest that laminin-8, which may facilitate tumor invasion, contributes to tumor regrowth after therapy. Laminin-8 may be used as a predictor of tumor recurrence and patient survival and as a potential molecular target for glioma therapy. Cancer 2004. © 2004 American Cancer Society.

Gliomas are the most common brain tumors and are comprised of different forms, including astrocytomas and oligodendrogliomas.1 The average survival time for patients with low-grade astrocytoma or oligodendroglioma is 6–8 years. Survival time decreases to 3 years for patients with anaplastic astrocytoma and drops to 12–18 months for patients with Grade 4 astrocytoma or glioblastoma multiforme (GBM). The majority of GBMs are highly invasive and rapidly recur at the site of the primary tumor. Oligodendroglioma is the third most common glioma type following GBM and astrocytoma, and is markedly different from these tumors in terms of survival rate (60% for oligodendroglioma vs. 5.5% for glioblastoma 5 years postoperatively) and treatment regimens.2 Despite the growing amount of information regarding the molecular, biochemical, and morphologic characteristics of these gliomas, the success of their treatment remains limited.3

Basement membranes (BMs) are specialized structures of the extracellular matrix (ECM) and play an important role in tumor progression because they can act as barriers against invasion or migration substrata for tumor cells, and/or form components of tumor blood vessels. Proteolytic degradation of ECM mediated by matrix metalloproteinases (MMPs) and plasminogen activators overexpressed by glioma cells may facilitate tumor cell migration and invasion.4 MMP-2 and MMP-9 are reportedly elevated in patients with gliomas and have been implicated in glioma invasion. They belong to the gelatinase family and are capable of degrading different BM components.5, 6 Along with the ECM proteolysis necessary for invasion, human gliomas demonstrate increased expression of the ECM and BM components associated with tumor vasculature and tumor stroma, including fibronectin, laminins, Types III and IV collagen, vitronectin, and tenascin-C.7–17

The major constituents of normal brain microvascular BMs are Type IV collagens, nidogens, perlecan, and the glycoproteins of the laminin family.15, 17, 18 The BMs of brain capillaries have a complex structure and are produced by both endothelial and glial cells.17

In the normal brain, the endothelial cells contribute laminins containing α4 and α5 chains to these BMs, whereas glial cells synthesize laminins containing α1 and α2 chains.17 In capillary BMs of normal human brain, we recently observed very weak expression of the α4 chain-containing laminin-9. It is interesting to note that during progression of human gliomas to GBM, the expression of capillary BM laminins containing the α4 chain dramatically increases and switches from the predominant laminin-9 (α4β2γ1) to the predominant laminin-8 (α4β1γ1).19 Laminin-8 and its receptor, integrin α6β1, appear to be important for the functioning of endothelial cell BMs, and have been implicated in the maintenance of the blood-brain barrier.20, 21 Some cultured glioma cell lines also can produce α4-containing laminins, mainly laminin-8.20, 22 Laminin-8 is believed to play a role in cell migration during development, wound healing, and angiogenesis,21, 23 while recent data also implicate laminin-8 in the invasion of GBM cells.22

Understanding the molecular mechanisms involved in the development of different gliomas and their recurrence is crucial to the development of more effective therapeutic approaches, predicting responses to treatment, and improving survival rates. Therefore, it is important to identify new biologic markers of tumor recurrence and invading glial tumors. The present study investigated whether a correlation exists between laminin-8 expression in GBM and time to recurrence and the survival of GBM patients.

MATERIALS AND METHODS

Tissue Samples

Fresh brain tissue samples were obtained from the Department of Pathology at Cedars-Sinai Medical Center with the approval of the Institutional Review Board for the Protection of Human Subjects (IRB #3637-3 from September 2, 2003). Immediately after surgery, tissues were frozen in liquid nitrogen and stored at -80 °C. Each sample was morphologically evaluated according to Daumas-Duport et al. and the World Health Organization (WHO) classifications of brain tumors,24, 25 before immunostaining and protein extraction for Western blot analysis.

Immunofluorescent Analysis

Sixty-seven tissue samples were used: 37 primary glial tumors including 23 GBMs, 6 Grade 2 astrocytomas, 13 low-grade glial tumors (7 Grade 1/2 astrocytomas, 2 oligoastrodendrogliomas, and 4 oligodendrogliomas), 11 glial tumor-adjacent tissues, 12 meningiomas, and 5 normal brains. In 16 cases, laminin α4, β1, and β2 chains were studied by both immunohistochemistry and Western blot analysis. Snap-frozen tissue samples were embedded in optimal cutting temperature (OCT) compound for cryosectioning. Eight-micrometer cryostat sections were processed for indirect immunofluorescence as described previously.19, 26 Well characterized monoclonal antibodies (MoAbs) were used to laminin chains β1 (clone LT3; Upstate Biotechnology, Lake Placid, NY), β2 (clone C4; Developmental Hybridoma Bank, Iowa City, IA), and γ1 (clone A5; gift from Dr. A. Ljubimov at Cedars-Sinai Medical Center).19 Rabbit polyclonal antibodies to the laminin α4 chain were #37718 and α4IIIa.27 In addition, a laminin α4 chain MoAb, FC10,28 was used in some pilot experiments. The different antibodies to the laminin α4 chain gave very similar results. Secondary cross-species absorbed fluorescein-conjugated and rhodamine-conjugated goat antimouse, antirat, and antirabbit antibodies, were obtained from Chemicon International (Temecula, CA). Polyclonal rabbit antibody to von Willebrand factor/factor VIII-related antigen (vWF) was used to visualize blood vessels and was obtained from Dako Corporation (Carpinteria, CA).

Routine specificity controls (without primary or secondary antibodies) were negative. MoAbs were used as straight hybridoma supernatant fluids or at 10–20 μg/mL when purified, and polyclonal antibodies were used at 20–30 μg/mL. At least two independent experiments were performed for each marker, with identical results. Sections were viewed and photographed with an Olympus BH-40 fluorescence microscope (Olympus America, Melville, NY) that was equipped with a MagnaFire digital camera (Optronics, Goleta, CA).

Staining intensity was graded as follows: -: no staining; +: weak staining; ++: distinct staining; +++: bright staining; and ++++: very strong staining. Some vessels in the same sample were considered to be in one staining category and some were considered to be in another staining category (Table 1).

Table 1. Laminin Isoform Expression in Normal, Benign, and Malignant Brain Tissues. Immunohistochemical Study
Sample no.Age (yrs)GenderDiagnosisRecurrent development / survival (mos)Laminin expression 8/9aα4bβ1β2γ1
  • M: male; GBM: glioblastoma multiforme; +++: bright staining; +/−: weak or no staining; +: weak staining; −: no staining; F: female; ++: distinct staining; ++++: very strong staining; NA: not available.

  • a

    The predominant isoform is shown.

  • b

    The slash between staining readings indicates that some vessels in the same sample were in one category whereas others were in another.

  • c

    Tissue was obtained from the same patient.

1645MGBM3/38+++++++/−+++
2258MGBM3.8/16.58++++++++/−+++
39c38MGBM4.16/178++++++++++
39, adjacent38MHistologically normal 8+++++++++
4259FGBM5.5/5.58+++++++++++
4570MGBM2.8/18.58+++++++++++++
49c47MGBM3.4/68++++++++++
49, adjacent47MHistologically normal 9+/+++++++++
5061MGBM8.6/10.59++/+++++++++
5179MGBM7/79+++++++++
47c57MGBM9/209+++++++++
47, adjacent57MFocal invasion 9++++/++++++++
5457MGBM5.2/168++++++++++++
7358FGBM6/158+++++++++++
86c55MGBM17/248++++++++++
86, adjacent55MGBM Low 9++/−++++++
9239MGBM16/229++++++++
93c43FGBM15.5/189++++/+++++++
93, adjacent43FGBM 9+++++++++
10053MGBM3/6.58+++++++/−+++
102c66MGBM3/58+++++++++++
102, adjacent66MGBM 9+++++++++++
10964FGBM4/109+++/++−/++++/+++++
11952MGBM4/NA8++++++++++++
128c30MGBM2/98+++++−/++++
128, adjacent30MGBM 9++++++++
6452FGBM6.8/98++++++++++
99c43FGBM14.3/228++++/+++/−+++
99, adjacent43FGBM 9+++/−+++++
6326MGBM3.5/128++++++++++
10455MGBM2.7/188+++++++++++
7944FAstrocytoma, Grade 3 9++++++++++
10722MAnaplastic astrocytoma, Grade 3 9++++++++++
10520MAnaplastic astrocytoma, Grade 3 Low 8++++++++
4127MAstrocytoma, Grade 3 9+/++++++++
9445FAstrocytoma, Grade 2 9++−/++++++
12932FAnaplastic oligoastrocytoma, Grade 3 9++/++++++++++
84c50FAnaplastic astrocytoma, Grade 20/39+/−+/−++++++
84, adjacent50FAnaplastic astrocytoma, Grade 2 9++++++++
4828MAstrocytoma, Grade 2 9++++++++
5332MAstrocytoma, Grade 2 9++++++++++
65c33MOligoastrodendroglioma, Grade 2 9++++++++++
65, adjacent33MOligoastrodendroglioma, Grade 2 Low 9+/+++++++++
6221MPilocytic astrocytoma, Grade 1 9++−/++++++
11331MLow-grade glioma, Grade 1–2 9+++++++++
10641MAnaplastic oligodendroglioma, Grade 3 Low 9+/+++++++
7828MAnaplastic oligodendroglioma, Grade 2 9+++/−+++++
108c28MOligodendroglioma, Grade 2 9+++−/++++++++
108, adjacent28MOligodendroglioma, Grade 2 9++−/+++++++
12239MOligodendroglioma, Grade 2 Low 9+/−+/−−/+++++
3553FMeningioma Unclear++/++++++
3846MMeningioma Unclear−/++/++++++
9044FMeningioma 9+++++++++
10139FMeningioma 9++/−+++++
11557MMeningioma 9++++/+++++++
11751MMeningioma 9+++++++
9655FMeningioma 9+++/−++++++
9737FMeningioma Unclear+++++++
9545MOlfactory meningioma, Grade 1 9++++/−+++++++
11262FAnaplastic malignant meningioma, Grade 3 8+++++++++++++
10362FAtypical meningioma Low 9+−/++++
11171FAtypical meningioma, Grade 2 9+++/+++++++++
6643FGliosis +/−−/+−/++++
13830MNormal brain tissue +/−−/++/+++++
4044FNormal brain tissue −/+++++++
4447MNormal brain tissue −/+++++++
4657MNormal brain tissue −/++++++

Western Blot Analysis

To obtain more quantitative data regarding protein expression, Western blot analysis of tissue extracts was performed using 16 human brain tissue samples (4 normal brains from trauma patients, 4 Grade 1/2 astrocytomas, 3 Grade 3 astrocytomas, and 5 GBMs [Grade IV astrocytomas]). The T98G glioblastoma cell line, which is known to express laminin-8 but not laminin-9,20 was used as a positive control.

Before sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), samples were solubilized in Laemmli reducing sample buffer with 1% SDS and 0.5 mM of dithiothreitol, and protein content was determined and normalized by the BCA method (Pierce Chemical Co., Rockford, IL). Equal amounts of protein were applied to each lane of Tris-acetate 8–12% gradient polyacrylamide gels (Invitrogen, Carlsbad, CA) and run under reducing conditions. Additional loading normalization was examined using the β-actin monoclonal antibody (clone AC-15; Sigma Aldrich Company, St. Louis, MO) staining. The gels were blotted onto nitrocellulose membranes (Invitrogen). The membranes were probed with antibodies followed by chemiluminescent detection using the ECL kit (Amersham, Buckinghamshire, UK). MoAbs were used to laminin chains α4 (clone 8B1220), β1 (LT3), and β2 (C4), and to β-actin (AC-15). The intensities of the bands of interest were determined using the AlphaImager™ 2000 densitometer (Alpha Innotech, Inc., San Leandro, CA) and expressed relative to the β-actin band intensities for the same specimens. The background was subtracted from each reading before the intensity ratios were determined.

Statistical Analysis

The interval between recurrence and survival were compared statistically for individual patient using laminin isoform expression as determined by immunohistochemistry using the Kaplan–Meier method (Prism4 program; GraphPad Software, San Diego, CA). The expression ratios of α4, β1, and β2 laminin chains relative to β-actin expression as revealed by Western blot analysis with normal brain and low-grade and high-grade tumor tissues were compared using ANOVA (Prism4 program). With both methods, a P < 0.05 was considered statistically significant.

RESULTS

Immunohistochemistry

We have previously shown that in brain tumors only two laminin isoforms containing the laminin α4 chain (laminin-8 and laminin-9) were overexpressed in microvascular BMs compared with normal brain tissue. All other isoforms containing laminin α1, α2, α3, and α5 chains were reported to be unchanged.19 Therefore, this expanded study focused on the laminin-8 (α4β1γ1) and laminin-9 (α4β2γ1) isoforms. Staining for the laminin α4, β1, and β2 chains was observed for the most part in blood vessel walls, although scattered parenchymal staining also was observed in some cases. Staining for the α4 chain was found to be weak in normal brain tissue (Fig. 1) and meningioma (data not shown). The staining was found to be stronger in all 13 low-grade (Grade 1/2) gliomas (Fig. 1) (Table 1), and was found to be significantly stronger in Grade 3 astrocytomas. In all 23 GBMs and the 8 GBM-adjacent tissues, the blood vessels generally demonstrated the strongest staining for the laminin α4 chain (Table 1).

Figure 1.

Double immunofluorescent staining of brain tissues for laminin-8 (α4β1γ1) and laminin-9 (α4β2γ1) chains, and for von Willebrand factor (vWF) specific for endothelial cells. N: normal brain, in which microvessels are positive for vWF, and α4 and β1 laminin chains are barely visible. At the same time, the β2 laminin chain is prominent in vessel walls that are positive for vWF. This pattern is compatible with small amounts of laminin-9. AS II: Grade 2 astrocytoma demonstrating stronger staining for the laminin α4 chain in brain microvessels. The expression of the laminin β1 chain is higher than in normal brain tissue and β2 remains strong. This pattern is compatible with a predominance of laminin-9. GBM: glioblastoma multiforme with very bright staining of α4 and β1 laminin chains but very weak β2 chain is noted in the brain vessels. This pattern is compatible with a predominance of laminin-8.

The laminin β1 chain was found to be weak in normal brain tissue, benign meningiomas, and low-grade gliomas. In the majority of these cases, distinct to strong staining for the laminin β2 chain was noted (Fig. 1), a finding that is compatible with the predominant expression of laminin-9. However, in two of the six Grade 3 astrocytomas, in 17 of 23 GBMs, and in 3 of 8 GBM-adjacent tissues, β1 chain staining was found to be stronger than in low-grade gliomas and normal brain tissue (Table 1). At the same time, laminin β2 chain staining was often markedly weaker than β1 staining in these cases, a finding that is compatible with laminin-8 up-regulation in approximately 67% of GBM cases (Fig. 1). Thus, normal brain tissue and low-grade gliomas were found to express relatively low levels of laminin-9. Laminin-8 expression appeared to be elevated in some Grade 3 astrocytomas, achieving peak expression in GBMs. The laminin γ1 chain antibody was found to brightly stain blood vessel walls in normal brain tissue and in all tumors studied (Table 1).

Western blot analysis was performed to confirm the changes in laminin expression suggested by immunofluorescence staining and to quantify any such changes. Equal amounts of protein per lane were analyzed for different tissue samples and each blot also was probed for β-actin. Sixteen samples of tumors of different grades and normal brain tissues were randomly selected from the same human brain tissue samples that were studied by immunohistochemistry. As shown in Figure 2, laminin-8 chains were expressed in all five GBM samples and in two of three Grade 3 astrocytomas. Grade 1/2 astrocytomas and normal brain tissues were found to be largely negative for laminin-8 expression, but expressed very low levels of laminin-9. Laminin-9 was found to be present in all three Grade 3 astrocytomas, two of which also were found to express laminin-8 (presumably, at a transitional stage before GBM).

Figure 2.

Western blot analysis of laminin isoforms in normal and malignant brain tissues. The laminin α4 chain migrated at 200 kilodaltons (kD), the laminin β1 chain migrated at 230 kD, the laminin β2 chain migrated at 190 kD, and β-actin migrated at 47 kD. Laminin-8 was expressed in all five cases of glioblastoma multiforme (GBM) and in two of three Grade 3 astrocytomas. All Grade 1-2 astrocytomas and normal brain tissues were found to be largely negative for the expression of laminin-8. Low-grade astrocytomas expressed low levels of laminin-9. Laminin-9 was present in all Grade 3 astrocytomas, two of which expressed laminin-8 at the same time (presumably, a transitional stage before the development of GBM).

The analysis of Western blots by quantitative densitometry (Table 2) confirmed that GBM specimens had significantly higher levels of the α4 laminin chain compared with normal brain tissue (P < 0.006) and Grade 1/2 gliomas (P < 0.012). The same was true for the laminin β1 chain (P < 0.005 and P < 0.007, respectively). Moreover, Grade 3 gliomas also were found to demonstrate significant overexpression of the laminin β1 chain compared with normal brain tissue and Grade 1/2 gliomas (P < 0.04 and P < 0.045, respectively). Laminin β2 chain levels were found to be significantly reduced in Grade 1/2 and Grade 4 gliomas compared with normal brain tissue (P < 0.02), with some nonsignificant reductions noted in Grade 3 gliomas (P = 0.0877).

Table 2. Expression Levels of α4, β1, and β2 Laminin Chains versus β-Actin in Different Sample Groups Obtained by Western Blot Analysis
Sampleα4β1β2
  • GBM: glioblastoma multiforme.

  • Data are expressed as the mean ± the standard error of the mean.

  • a

    P < 0.05 versus Grade 4.

  • b

    P < 0.05 versus Grade 3.

  • c

    P < 0.05 versus Grade 1/2.

Normal brain9.96 ± 1.63a0.48 ± 0.38ab54.21 ± 17.14ac
Grade 1/2 glioma14.80 ± 5.26a2.23 ± 0.92ab15.58 ± 5.15
Grade 3 glioma22.05 ± 9.2129.84 ± 9.0125.37 ± 11.93
Grade 4 glioma (GBM)46.99 ± 10.6737.29 ± 11.2816.15 ± 3.56
Positive control47.9246.210

The expression of laminin-8 versus laminin-9 also was compared statistically between Western blot analysis and immunohistochemistry using the number of cases with a predominant expression of a respective laminin isoform. Using immunohistochemistry (Table 1), laminin-8 was found to be overexpressed in 17 of 23 GBM cases (74%) compared with none of the 5 normal brain cases (P < 0.005 using the Fisher exact test). Using semiquantitative Western blot analysis (Fig. 2), laminin-8 overexpression was noted in 4 of 5 GBM cases (80%) compared with none of the 4 cases of normal brain tissue (P < 0.05). At the same time, there was no significant difference noted between the number of GBM cases with predominant expression of laminin-8 revealed by either method (P = 1.000). Overall, data from the Western blot analysis demonstrated significantly increased levels of laminin-8 and decreased laminin-9 levels during progression of brain gliomas, as suggested by the immunofluorescent results.

In the 23 patients with GBM, laminin-8 expression was compared with regard to tumor recurrence (similar to what we previously published for 9 GBM patients19) and patient survival time. In all these patients, the tumors had recurred by the time of last follow-up (September 1, 2003 [Table 1]), and death records were available for 22 patients. The death record was not found for one patient who underwent surgery twice at the Neurosurgical Institute. Clinically, all 17 patients with GBM who demonstrated high laminin-8 levels developed a tumor recurrence at a mean of 3.92 months after surgery, whereas 6 patients with a high expression of laminin-9 developed disease recurrence at a mean of 11.43 months after surgery. This difference was found to be highly significant (P < 0.0002) (Fig. 3). Therefore, the new statistical data with an expanded number of patients confirmed our previous study concerning nine GBM patients.19 The mean survival time for 16 GBM patients whose tumors predominantly expressed laminin-8 was 11.2 months, whereas it was 16.7 months for the 6 GBM patients whose tumors expressed laminin-9 (P < 0.015) (Fig. 3).

Figure 3.

Laminin-8 was found to be associated with a decreased time to the development of recurrent tumor and decreased survival in patients with glioblastoma multiforme compared with laminin-9. Statistical analysis was performed using the Kaplan–Meier method. Solid line: laminin-8; dashed line: laminin-9.

The majority of Grade 4 gliomas (GBM) had increased levels of laminin-8 (α4β1γ1) in blood vessel walls, whereas in low-grade tumors laminin-9 (α4β2γ1) predominated. GBMs that predominantly expressed laminin-8 were associated with a shorter time to tumor recurrence and a shorter patient survival compared with GBMs that predominantly expressed laminin-9.

DISCUSSION

Brain tumors account for millions of dollars every year in hospital stays, surgery, radiation, and chemotherapy. The majority of these tumors are not cured by surgery, radiation, or chemotherapy and approximately 60% of brain tumors are GBMs that kill patients in approximately 12–18 months. Therefore, the successful management of brain tumors requires constant research targeted toward the development of more efficient treatments and prognostic approaches.

Success in the diagnosis, treatment, and prognosis of brain gliomas depends largely on awareness of the specific gene expression differences between gliomas and normal brain tissue. Therefore, the search for new reliable markers of gliomas becomes the first priority. We utilized gene array technology to identify complex profiles of gene expression in gliomas compared with normal brain tissue and to date have identified only two genes, epidermal growth factor receptor (EGFR) and laminin α4 chain, that are overexpressed in both high-grade and low-grade gliomas.19

For most tumors, including brain gliomas, the known molecular changes are of rather limited use for diagnosis, prognosis, or treatment.2 To our knowledge to date, determination of the differential expression of markers in gliomas has not significantly altered existing therapeutic approaches, the treatment success rate, or the prediction of disease outcome.2, 29 Determining specific brain tumor markers, especially for low-grade gliomas (which often progress to become highly invasive, extensively vascularized, and rapidly recurring GBMs), therefore is extremely important.

A number of genes and proteins have been identified as having altered expression in glial tumors. Most recently, some of the markers have been tested for diagnostic and prognostic purposes. These markers include EGFR, tenascin-C, the bcl-2 family of antiapoptotic proteins, survivin, Rho proteins, p53, and vascular endothelial growth factor (VEGF) and its receptors.9, 30–34 The results of these tests to date have been controversial. Some reports suggest that specific proteins (e.g., Rho, VEGF, and EGFR) could be used to discriminate between low-grade and high-grade gliomas. The increased expression of some proteins (e.g., EGFR, tenascin-C, and survivin) was found to be correlated with shorter patient survival.15, 16, 35 At the same time, some of these markers (e.g., VEGF) are up-regulated in a variety of tumors and are not glioma-specific. Other markers (e.g., the bcl-2 family of proteins and EGFR) did not appear to be correlated significantly with survival when large groups of patients were studied.35–37 Therefore, before being considered as diagnostic and/or prognostic tools, these and other candidate markers must be analyzed thoroughly in well controlled studies involving significant numbers of patients and tumors of different grades. Unfortunately, to our knowledge, the vast majority of existing markers of tumor invasion and disease progression have not yet been subjected to rigorous preclinical and clinical testing.

Our laboratory has previously identified a novel promising glioma marker, laminin-8, which is overexpressed in GBMs compared with normal brain tissue. Laminin-8 and laminin-9 are constituents of vascular BMs, but in normal brain capillaries the endothelium apparently can produce only low levels of laminin-9. In gliomas, especially GBMs, the expression of these laminins markedly increases and laminin-8 predominates. This may be related to the neovascularization that is a hallmark of brain glial tumors. Moreover, we recently took a step further in establishing the role of laminin-8 in glioma invasion and in developing means with which to use apply this finding for therapeutic purposes. Antisense oligos blocking laminin-8 expression could efficiently reduce glioma invasion in vitro.22 Therefore, laminin-8 also may be a promising target for future glioma therapy using a safe and specific antisense approach. If tumor cells in vivo produce laminin-8 (as we have confirmed for cultured glioma cell lines) or induce it in endothelial cells, this laminin may contribute to local invasion.22 Laminin-8 is rather poorly adhesive for cells but appears to support cell migration,20 which is necessary for local tumor invasion. By such a mechanism, overexpression of this protein may be important for tumor invasion and metastasis. Clinically, this would translate into a faster rate of GBM recurrence and decreased patient survival, as suggested by the correlative data from the current study.

Enhanced expression of laminin-8 in high-grade compared with low-grade gliomas suggests its involvement in tumor progression. Overexpression of laminin-8 in tumor-adjacent tissues may facilitate the spread of microinvasive glioma foci not removed by surgery or available therapy and may lead to disease recurrence. This may explain why laminin-8 may be predictive of glioma recurrence. Given these patterns of laminin-8 (and to a lesser extent, laminin-9) expression, we hypothesize that laminin-8 plays a role in glioma progression and tumor recurrence, and that inhibiting its expression may impact glioma invasion and recurrence.

Laminins are the major constituents of blood vessel BMs. A switch from laminin-9 to laminin-8 expression, with its gradual increase from a low level of expression in low-grade glial tumors (Grade 1/2) to a moderate level of expression in Grade 3 gliomas to a significantly high level of expression in 74% of GBMs (Grade 4), may be associated with neovascularization and therefore contribute to tumor aggressiveness. Overall, overexpression of laminin-8 in GBMs, together with factors promoting tumor growth (e.g., EGFR), might be an important prognostic marker for predicting the time to recurrence and survival time in patients with GBM. As a factor that may play a role in the process of invasiveness in human gliomas as well as in their progression, laminin-8 appears to be a promising target for new therapeutic approaches, possibly in combination with other tumor markers or standard chemotherapy and radiation.

Acknowledgements

The authors are grateful to Dr. Alexander V. Ljubimov (Cedars-Sinai Medical Center) for helpful suggestions and critical reading of the article, and to Dr. Ismo Virtanen (Institute of Biomedicine/Anatomy, University of Helsinki, Helsinki, Finland) for the gift of the laminin α4 chain monoclonal antibody FC10.

Ancillary