Evaluation of sensitivity and specificity of doublecortin ...

Brain Tumor Pathol (2008) 25:1–7 DOI 10.1007/s10014-007-0225-1

© The Japan Society of Brain Tumor Pathology 2008

ORIGINAL ARTICLE Kenta Masui · Shin-ya Mawatari · Satoshi O. Suzuki Toru Iwaki

Evaluation of sensitivity and specificity of doublecortin immunostaining for the detection of infiltrating glioma cells

Received: September 4, 2007 / Accepted: October 29, 2007

Abstract Diffuse gliomas are highly infiltrative intracranial tumors, but there are few useful markers for detecting infiltrating glioma cells in the surrounding brain tissue. Doublecortin (DCX) is a microtubule-associated protein (MAP) that plays a crucial role in neuroblast migration. It was recently demonstrated that DCX is preferentially expressed in invasive gliomas. However, the sensitivity and specificity of DCX as a marker for infiltrating glioma cells have not been fully evaluated. We immunohistochemically analyzed the expression pattern of DCX in human gliomas and compared it with that of MAP-2e, another marker for infiltrating glioma cells. We found that DCX was expressed specifically in infiltrating gliomas, but not in reactive, existing glia. Not all our cases exhibited stronger immunoreactivity to DCX at the invasive margin than at the core mass. The level of DCX expression was more variable from case to case than that of MAP-2e. For the identification of infiltrating glioma cells, DCX was thus more specific than MAP2e whereas MAP-2e was more sensitive than DCX. DCX immunostaining would detect infiltrating low-grade glioma cells that are not efficiently labeled by proliferative markers. Taken together, DCX is applicable for the detection of individual infiltrating glioma cells when combined with other markers. Key words Doublecortin · MAP-2e · Glioma · Infiltrating tumor cells · Immunohistochemistry

Introduction Diffuse gliomas, including astrocytomas, oligodendrogliomas, and glioblastomas, are the most common type of primary intracranial tumor in humans. One of the cardinal

K. Masui (*) · S. Mawatari · S.O. Suzuki · T. Iwaki Department of Neuropathology, Neurological Institute, Graduate School of Medical Sciences, Kyushu University, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan Tel. +81-92-642-5537; Fax +81-92-642-5540 e-mail: [email protected]

features of diffuse gliomas is that tumor cells migrate out of the core mass to infiltrate the surrounding brain tissue, and individual tumor cells are often detectable as far as several centimeters away from the core lesion.1 In the practice of surgical pathology, pathologists are often required to evaluate the invasive front of gliomas. Moreover, there is sometimes a need to judge if a lesion is neoplastic or nonneoplastic when clinical data including radiologic findings do not meet a definite diagnosis. In many cases, however, it is difficult to distinguish infiltrating glioma cells from reactive, existing glial cells on the basis of their cellular morphology alone, especially in small biopsy samples. Therefore, it is challenging but crucial for pathologists to develop histological markers to identify infiltrating glioma cells among nontumor constituents at the tumor periphery. To date, there is no ideal marker for this purpose, in terms of specificity and sensitivity, because most markers for glioma cells are also markers for glia of astrocytic or oligodendroglial lineages; these markers include glial fibrillary acidic protein (GFAP), S-100 protein, vimentin, nestin, and NG2.2,3 Glioma cell invasion bears a striking resemblance to the robust migration of glial and neuronal cells during embryogenesis.4 Doublecortin (DCX) is a 40-kDa microtubuleassociated protein (MAP) involved in microtubule dynamics and stabilization. It is highly expressed in migrating neural precursor cells during the development of the cerebral cortex, and in the subventricular zone and rostral migratory stream from the embryonic period to adulthood.5–9 Mutations in the human DCX gene mapping to Xq22.3-q23 are responsible for X-linked lissencephaly and subcortical band heterotopia or “double cortex,”10,11 further supporting the idea that DCX plays a critical role in neuronal migration. A recent study by Daou et al.12 demonstrated that DCX was preferentially expressed in invasive brain tumors including gliomas, and that DCX immunostaining was stronger at the margin of the tumor than at the center. In their study, however, the sensitivity and specificity of DCX expression in infiltrating glioma cells were not sufficiently examined, especially at the invasive margin of the tumor. To evaluate the usefulness of DCX immunostaining for identifying infiltrating glioma cells in the surrounding brain

2

tissue, we examined the tissue distribution of DCX in a variety of glial tumors and compared the findings with the expression pattern of MAP-2e, a recently identified splicing variant of MAP-2 that has been shown to delineate glioma invasion into the adjacent brain tissue.13 We show that DCX is expressed specifically in infiltrating gliomas, but not in reactive astrocytes or microglia, while it is known that MAP-2e is expressed in a certain population of existing glia.13 DCX expression was noted both at the center of the tumor and at the margin regardless of tumor tissue type or grade, and some cases exhibited stronger immunoreactivity at the center than at the margin while others presented the inverse pattern. The level of DCX expression was more variable from case to case, even in the same tumor type and grade, than that of MAP-2e expression. Therefore, for the identification of infiltrating glioma cells, DCX was more specific than MAP-2e whereas MAP-2e was more sensitive than DCX. Because DCX was expressed in low- to high-grade gliomas, it could be an alternative marker for low-grade infiltrating glioma cells in place of proliferative markers such as Ki-67 (MIB-1) and proliferating cell nuclear antigen (PCNA), which label only a small fraction of low-grade glioma cells. These findings indicate that DCX can be of value for specifically detecting individual infiltrating glioma cells when combined with other markers, such as MAP-2e.

Materials and methods Tissue collection A total of 29 human glioma tissues and nonneoplastic adult human cerebral tissues from two epileptic patients were obtained at either surgery or autopsy. The tumor samples included 5 grade I gliomas [2 gangliogliomas (GGs), 3 pilocytic astrocytomas (PAs)], 8 grade II gliomas [5 diffuse astrocytomas (DAs), 2 oligodendrogliomas (OLs), 1 ependymoma (EP)], 5 grade III gliomas [4 anaplastic astrocytomas (AAs), 1 anaplastic oligodendroglioma (AOL)], and 11 grade IV gliomas [10 glioblastomas (GBMs), 1 gliosarcoma (GS)]. All specimens were fixed in 10% formalin and processed into paraffin sections. Sections were routinely stained with hematoxylin and eosin (H&E), and histopathological diagnoses were made at the Department of Neuropathology, Kyushu University, according to the WHO classification of tumors of the nervous system.14 Immunohistochemistry The antibodies used in this study include anti-DCX (C-18; Santa Cruz Biotechnology, Santa Cruz, CA, USA, goat polyclonal) used at a dilution of 1 : 150, anti-MAP-2e (13-10, Dr. Shafit-Zagardo, mouse monoclonal) used at 1 : 50, antiGFAP (Dako A/S, Glostrup, Denmark, rabbit polyclonal) used at 1 : 1000, and anti-Iba-1 (IBL, Gunma, Japan, rabbit polyclonal) used at 1 : 500.

All sections were deparaffinized in xylene and rehydrated in an ethanol gradient. Sections for DCX and MAP-2e staining were subjected to antigen retrieval by autoclaving in 0.01 M citrate buffer, pH 7.0 (for DCX) or distilled water (for MAP-2e). Endogenous peroxidase activity was blocked with 0.3% H2O2/methanol. The sections were then incubated with primary antibody at 4°C overnight. After rinsing, the sections were subjected to either the streptavidin–biotin complex method (for DCX, GFAP, and Iba-1) or the enhanced indirect immunoperoxidase method using Envision (DakoCytomation, Glostrup, Denmark) (for MAP-2e). Immunoreactivity was detected using 3,3′-diaminobenzidine (DAB), and sections were counterstained with hematoxylin. DCX and MAP-2e immunohistochemistry was performed in all specimens. To confirm the specificity of DCX expression in glioma cells, adjacent sections from epileptic cases were also immunostained with anti-DCX, anti-GFAP, and anti-Iba-1 antibodies. The percentage of DCX-positive cells in each case was semiquantitatively evaluated as ± (a few weakly positive cells), + (positive cells < 50%), and ++ (positive cells ≥ 50%). Double immunofluorescence To examine the immunoreactivity of the anti-DCX antibody to existing microglia, indirect double immunofluorescence was performed on sections of an anaplastic astrocytoma, using combinations of the aforementioned primary antibodies against DCX and Iba-1. Fluorescein isothiocyanate (FITC)- or Texas red-conjugated secondary antibodies (Southern Biotech, Birmingham, AL, USA) were used for detection. Sections were then examined under a confocal microscope (Radiance 2000; Bio-Rad, Hercules, CA, USA).

Results Specificity of the antibodies We first stained nonneoplastic adult human cerebral tissues with anti-DCX C-18 antibodies. Most cells in these tissues, including neurons, reactive astrocytes, and activated microglia, were DCX negative, whereas GFAP and Iba-1 immunostaining detected numerous reactive astrocytes and microglia, respectively. A few weakly positive neurons were detected in the hippocampus (data not shown). These results are consistent with previous data on DCX expression.5,8,9 To confirm the specificity of DCX immunostaining in glioma cells, we carried out double staining of sections with anaplastic astrocytoma labeled by anti-DCX and anti-Iba-1. No double-positive cells were detected in DCX/Iba1 doublelabeled sections (Fig. 1). This experiment further confirmed that DCX protein was exclusively localized to the neoplastic cells in the human glioma tissues.

3 Table 1. Semiquantitative analysis of doublecortin (DCX) immunostaining in human glioma tissues

Grade I (n = 5) Grade II (n = 8) Grade III (n = 5) Grade IV (n = 11)b

±

+

++

2 1 0 2

3a 4 2 4

0 3 3 5

The percentage of DCX-positive tumor cells in each case was evaluated as ± (a few weakly positive cells), + (positive cells < 50%), and ++ (positive cells ≥ 50%) a Ganglion cell components in gangliogliomas (GGs) b This group includes two autopsy cases

population of tumor cells at the invasive margin than at the center of the tumor (Fig. 3E–H). Expression levels of DCX and MAP-2e in human glioma tissues

Fig. 1. Doublecortin (DCX) and Iba-1 immunostaining in human glial tumors. Double immunofluorescence for DCX (red) and Iba-1 (green) on a section of anaplastic astrocytoma reveals no double-positive cells. Bar 20 μm

Expression patterns of DCX and MAP-2e in human glioma tissues We then stained human diffuse glioma tissue sections, including DAs, OLs, EPs, AAs, and GBMs, with the antiDCX antibody, and found that all these tumors were immunopositive for DCX, at least in certain populations of tumor cells (Figs. 2C–F, 3C,D). Localized glial tumors, such as PAs and GGs, showed a low level of DCX immunoreactivity although some neuronal components in GGs exhibited strong DCX positivity (Fig. 2A,B). DCX was localized to the soma and processes of the tumor cells, delineating their various morphologies including bipolar tumor cells in PAs (Fig. 2B), glioma cells with fine, fibrillary processes in low-grade DAs (Fig. 2C), and those with plump, bizarre cytoplasm in high-grade astrocytomas (Figs. 2F, 3C), as well as infiltrating glioma cells exhibiting migratory morphology with unipolar or bipolar processes in diffuse gliomas of all grades (Fig. 3D). In GGs, PAs, OLs, and EPs, tumor cells tended to show weaker staining for both DCX and MAP-2e than astrocytic tumors (Fig. 2A,B,D,E). We then examined DCX distribution within individual tumors, comparing the central region of the tumor with the invasive margin. We found that tumors expressed DCX both at the center of the tumor and at the invasive margin, and some tumors exhibited stronger immunoreactivity at the center than at the margin, whereas the staining at the invasive margin was more intense than that at the center in other cases (Fig. 3A–D). These findings are independent of tumor tissue type or grading. By contrast, MAP-2e immunostaining was clearly more intense and seen in a greater

For semiquantitative analysis of the DCX immunostaining, we scored the staining intensities as ± (a few weakly positive cells), + (positive cells < 50%) and ++ (positive cells ≥ 50%) (Table 1). Staining intensity was generally lower in grade I localized gliomas than in diffuse gliomas, as previously reported.12 However, within diffuse gliomas, staining intensity was not correlated with either tumor type or grade, but rather varied from sample to sample, even in the same type and grade of tumor, whereas MAP-2e-immunostaining showed relatively constant staining intensity regardless of the tumor type and grading (Fig. 4A–D).

Discussion In this study, all tumor samples, including localized gliomas (PAs and GGs) and diffuse gliomas (DAs, OLs, AAs, and GBMs), were immunopositive for DCX to various degrees. Additionally, most nonneoplastic cells, including neurons, reactive astrocytes, and activated microglia, were DCX negative in both human nonneoplastic brain tissues and glioma tissues. Thus, these results indicate that immunostaining for DCX is useful for the diagnosis of gliomas and identification of infiltrating glioma cells against a background of nonneoplastic cell populations. A certain analogy exists between glioma cell invasion and the migration pattern of normal neural progenitor cells during development. Previous studies using time-lapse video imaging of progenitor cells15,16 and glioma cells17 migrating in living brain slices have revealed remarkable similarity in the morphology and dynamics of migrating progenitors and glioma cells, in which protrusion of a leading process precedes nuclear translocation. This migratory behavior supports a role of microtubules and associated proteins in active glioma migration and raises their potentiality as markers for migrating glioma cells. One such molecule is the MAP-2 isoform containing exon 13

4 Fig. 2. Expression of DCX in human glioma tissues. DCX immunostaining for ganglioglioma (A), pilocytic astrocytoma (B), diffuse astrocytoma (C), oligodendroglioma (D), ependymoma (E), and anaplastic astrocytoma (F). All samples are immunopositive for DCX; diffusely infiltrating astrocytic tumors (C, F) show strong staining, and noninvasive, grade I tumors (A, B) and nonastrocytic gliomas (D, E) exhibited only weak staining. Some neuronal components in gangliogliomas show strong poitivity for DCX (arrowhead, A). Arrows denote the DCXpositive cells (A–F). Bars 15 μm

A

B

C

D

E

F

(MAP-2e), which is highly expressed in human fetal brain and a population of adult glial progenitors.18 This molecule has been demonstrated to delineate glioma invasion in the adjacent gray and white matter and appears to be a useful marker for examining infiltrating glioma cells.13 In the present study, we compared DCX expression patterns with those of this known marker for infiltrating glioma cells, MAP-2e, to corroborate the value of DCX as a marker for infiltrating glioma cells. A recent study suggested that DCX immunostaining was more intense at the margin of the tumor than in the main mass.12 Although we confirmed that DCX expression was generally weaker in less invasive, localized gliomas compared with diffusely infiltrating gliomas, as Daou et al.12 demonstrated, we found that in individual diffuse gliomas, some showed stronger immunoreactivity at the center of the tumor than that at the invasive margin while others presented the opposite pattern. This result was different from the finding that MAP-2e staining was consistently more intense at the margins of tumors than in the main tumor. Thus, the correlation between DCX

expression and invasiveness was not as clear in diffusely infiltrating gliomas as previously reported.12 It has been recently reported that forced expression of DCX in DCXdeficient glioblastoma cell lines significantly suppressed their growth, suggesting that in glioma cells, DCX plays a role not only in migration but also in tumor suppression.19 Together with our findings, the expression of DCX in glioma cells thus may not necessarily reflect their infiltrative potential. Although MAP-2e is expressed in a small population of adult glial progenitor cells,13 we found almost no DCX-positive nonneoplastic glial cells in the adult brain, including reactive or nonreactive astrocytes, oligodendrocytes, and microglia. Therefore, generally speaking, for the identification of infiltrating glioma cells, MAP-2e was more sensitive than DCX and DCX was more specific than MAP-2e. Moreover, in some cases, DCX was expressed in a different population of tumor cells from MAP-2e-positive tumor cells, when the distributions of these markers were compared in adjacent sections (data not shown).

5 Fig. 3. Distribution of DCX and MAP-2e in human diffuse glioma tissues. DCX immunostaining is stronger at the center (A) than the margin (B) in an anaplastic astrocytoma, whereas the center (C) and the margin (D) show no significant difference in staining intensity in a glioblastoma. MAP-2e immunostaining is more intense in the margin (F, H) than the center (E, G) in both anaplastic astrocytoma (E, F) and glioblastoma (G, H). Arrows denote the DCXpositive (B–D) or MAP-2epositve (E–G) cells. Bars 15 μm

A

B

C

D

E

F

G

H

To examine whether the DCX expression level correlates with the histological grading, we performed a semiquantitative analysis using scores for staining intensities and the percentage of DCX-positive cells. We found that the low-grade and high-grade diffusely infiltrating gliomas were not significantly different from each other with respect to the DCX expression level. DCX immunoreactivity varied from case to case, regardless of the tumor grade. From these

findings and the previous report,19 DCX expression might be relevant to common intracellular mechanisms of glioma infiltration and proliferation, from low to high grade. Presently, proliferative markers such as Ki-67 and PCNA are often used to detect infiltrating glioma cells in the practice of surgical pathology. Low-grade diffuse gliomas including diffuse astrocytomas actively infiltrate into the surrounding brain tissue, but only a small fraction of the

6 Fig. 4. Expression levels of DCX and MAP-2e in human glioma tissues: DCX (A, B) and MAP-2e (C, D) immunostaining in two cases of diffuse astrocytomas (A and C for one case, B and D for another). Staining intensity of DCX varied from sample to sample even in the same type or grade of the tumor whereas that of MAP-2e was relatively constant. Arrows denote the DCX-positive cells (A, B). Bars 15 μm

A

B

C

D

tumor cells are labeled by these proliferative markers. Furthermore, reactive glial cells frequently exhibit the proliferative activity. Because not only high-grade gliomas but also low-grade gliomas express DCX, DCX can be an effective marker for exclusively detecting infiltrative glioma cells as well as MAP-2e. Together, our results demonstrate that DCX is specifically localized to tumor cells in human glioma tissues, and is expressed at various levels both at the tumor margin and at the center, regardless of tumor tissue type or grading. Despite the case-to-case variation in staining intensity and positive ratio, DCX can be a useful marker for detecting infiltrating tumor cells in human gliomas in clinical practice when combined with other markers including MAP-2e. Acknowledgments The authors thank Bridget Shafit-Zagardo (Montefiore Hospital) for generously providing the anti-MAP-2e antibody.

References 1. Giese A, Bjerkvig R, Berens ME, et al (2003) Cost of migration: invasion of malignant gliomas and implications for treatment. J Clin Oncol 21:1624–1636 2. The Committee of Brain Tumor Registry of Japan, The Japanese Pathological Society (2002) General rules for clinical and pathological studies on brain tumors, 2nd edn. Japanese Pathological Society, Tokyo, pp 25–31 3. Shoshan Y, Nishiyama A, Chang A, et al (1999) Expression of oligodendrocyte progenitor cell antigens by gliomas: implications for the histogenesis of brain tumors. Proc Natl Acad Sci U S A 96:10361–10366 4. Hatten ME (1999) Central nervous system neuronal migration. Annu Rev Neurosci 22:511–539

5. Brown JP, Couillard-Despres S, Cooper-Kuhn CM, et al (2003) Transient expression of doublecortin during adult neurogenesis. J Comp Neurol 467:1–10 6. Francis F, Koulakoff A, Boucher D, et al (1999) Doublecortin is a developmentally regulated, microtubule-associated protein expressed in migrating and differentiating neurons. Neuron 23:247–256 7. Gleeson JG, Lin PT, Flanagan LA, et al (1999) Doublecortin is a microtubule-associated protein and is expressed widely by migrating neurons. Neuron 23:257–271 8. Nacher J, Crespo C, McEwen BS (2001) Doublecortin expression in the adult rat telencephalon. Eur J Neurosci 14:629–644 9. Yang HK, Sundholm-Peters NL, Goings GE, et al (2004) Distribution of doublecortin expressing cells near the lateral ventricles in the adult mouse brain. J Neurosci Res 76:282–295 10. Gleeson JG, Allen KM, Fox JW, et al (1998) Doublecortin, a brainspecific gene mutated in human X-linked lissencephaly and double cortex syndrome, encodes a putative signaling protein. Cell 92:63–72 11. des Portes V, Pinard JM, Billuart P, et al (1998) A novel CNS gene required for neuronal migration and involved in X-linked subcortical laminar heterotopia and lissencephaly syndrome. Cell 92:51–61 12. Daou MC, Smith TW, Litofsky NS, et al (2005) Doublecortin is preferentially expressed in invasive human brain tumors. Acta Neuropathol (Berl) 110:472–480 13. Suzuki SO, Kitai R, Llena J, et al (2002) MAP-2e, a novel MAP-2 isoform, is expressed in gliomas and delineates tumor architecture and patterns of infiltration. J Neuropathol Exp Neurol 61: 403–412 14. Kleihues P, Cavenee WK (2000) World Health Organization classification of tumours of the nervous system. IARC/WHO, Lyon 15. Kakita A, Goldman JE (1999) Patterns and dynamics of SVZ cell migration in the postnatal forebrain: monitoring living progenitors in slice preparations. Neuron 23:461–472 16. Suzuki SO, Goldman JE (2003) Multiple cell populations in the early postnatal subventricular zone take distinct migratory pathways: a dynamic study of glial and neuronal progenitor migration. J Neurosci 23:4240–4250

7 17. Farin A, Suzuki SO, Weiker M, et al (2006) Transplanted glioma cells migrate and proliferate on host brain vasculature: a dynamic analysis. Glia 53:799–808 18. Shafit-Zagardo B, Davies P, Rockwood J, et al (2000) A novel MAP-2 isoform expressed early in human oligodendrocyte maturation. Glia 29:233–245

19. Santra M, Zhang X, Santra S, et al (2006) Ectopic doublecortin gene expression suppresses the malignant phenotype in glioblastoma cells. Cancer Res 66:11726–11735