WT1 Is Not a Reliable Marker to Distinguish Reactive from Neoplastic Astrocyte Populations in the Central Nervous System

Abstract

A diagnostic difficulty in neuropathology practice is distinguishing reactive from neoplastic astrocyte populations. This is particularly true in small biopsy samples that lack evidence of increased cellularity or mitotic activity, microvascular proliferation, or necrosis. We performed the current study to validate the previously reported finding that in the central nervous system, the expression of WT1 is limited to neoplastic astrocytes. We retrospectively studied WT1 protein expression by immunohistochemistry (IHC) in 100 formalin‐fixed, paraffin‐embedded brain tissue samples consisting of 3 normal control tissues, 44 cases of reactive gliosis, 49 gliomas and 4 lesions suspicious for glioma. In normal human cortex, WT1 staining was restricted to vascular endothelium. Most cases of reactive gliosis (82%) showed at least focal WT1 positivity, and analysis of specimens with electrode monitoring lesions showed an inverse relationship between WT1 expression intensity and the number of days from electrode placement to tissue resection. All glioma samples (100%) and all cases suspicious for glioma (100%) showed at least focal WT1 positivity. Our results likely differ from those in the prior report because of differences in tissue fixation and IHC methodology. Thus, our findings indicate that WT1 expression alone is not a reliable feature to distinguish reactive from neoplastic astrocytes.

BACKGROUND

A challenging diagnostic difficulty in neuropathology practice is distinguishing reactive from neoplastic astrocyte populations. This is particularly true in small biopsy samples that lack evidence of other malignant features such as hypercellularity, increased mitotic activity, microvascular proliferation or necrosis. In such cases, the differential diagnosis often includes reactive gliosis and diffusely infiltrating astrocytoma.

While a number of immunohistochemical markers, such as the tumor suppressor protein p53, have been proposed over the years to help distinguish reactive from neoplastic astrocyte populations (9), all have suffered from a lack of sensitivity, specificity or both. Another approach has involved the use of microdissection‐based genotyping to help distinguish reactive from neoplastic astrocyte populations (5). More recently, the utility of mutation analysis for isocitrate dehydrogenase 1 (IDH1) and isocitrate dehydrogenase 2 (IDH2) has been demonstrated as IDH1/2 mutations appear to be absent in nonneoplastic glial tissue (8). In addition, mutant IDH1‐specific analysis by immunohistochemistry (IHC) has been reported to reliably distinguish diffuse astrocytoma from astrocytosis (2).

Schittenhelm et al studied the expression of the Wilms' tumor associated protein (WT1), a zinc‐finger transcription factor, in a series of cases of reactive and neoplastic astrocyte populations (17). The authors report that WT1 expression is limited to neoplastic astrocyte populations, and they describe their experience with 48 cases of “reactive” gliosis in which no WT1 expression was identified.

Given the useful potential application of this finding in clinical neuropathology practice, we performed the current study in an attempt to validate these reported findings by testing the hypothesis that WT1 expression in the central nervous system is limited to neoplastic astrocyte populations.

METHODS

The study was performed with approval from the University of Virginia Institutional Review Board (HSR #14042).

Cases

We searched the University of Virginia departmental archive electronic database to locate formalin‐fixed paraffin‐embedded (FFPE) neurosurgical brain biopsy and resection specimens containing reactive and neoplastic astrocyte populations. We also searched the database to identify diagnostically challenging cases for which the determination of “reactive” vs. “neoplastic” was not clearly evident from review of the biopsy material.

For specimens containing reactive gliosis, we retrieved 44 cases with sufficient quantity of tissue for analysis from our departmental archives using search terms including “gliosis,”“hippocampal sclerosis,”“metastatic,”“meningioma with brain invasion,”“demyelinating,”“infarct,”“abscess” and “vascular malformation.” The cases showed gliotic areas associated with epilepsy resection specimens [11], metastatic carcinoma [10], normal pressure hydrocephalus [5], primary brain tumors [8], vascular malformations and hematomas [7], demyelinating disease and infarct [2], and intracranial abscess [1]. Among the epilepsy specimens, nine were from patients who had undergone prior electrode placement for pre‐resection seizure monitoring.

For specimens with neoplastic astrocyte populations, we retrieved 49 cases with sufficient quantity of tissue for analysis from our departmental archives using search terms including “astrocytoma,”“pilocytic astrocytoma,”“oligodendroglioma,”“anaplastic astrocytoma” and “glioblastoma.” The cases consisted of pilocytic astrocytomas [5], oligodendrogliomas [2], anaplastic astrocytomas [2], glioblastomas [38], giant cell glioblastoma [1] and gliosarcoma [1]. Of these cases, 40 consisted of glioblastoma and gliosarcoma samples from a previously constructed tissue microarray.

For specimens considered diagnostically challenging, we retrieved four specimens using the search terms “suggestive of” or “suspicious for” along with “glioma.”

We also retrieved three cases of normal‐appearing FFPE cortical brain tissue, resected as part of larger temporal lobe resections, to serve as normal control tissue. Each case in the study was initially reviewed by at least one neuropathologist, and one of the authors (T.D.B) re‐reviewed all original hematoxylin and eosin (H&E)‐stained sections to confirm each diagnosis.

Immunohistochemistry (IHC)

Tissue samples were initially fixed and routinely processed using 10% zinc formalin. If more than one paraffin block was available for a given case, the block with the largest amount of viable tissue of interest was selected for immunohistochemical testing. From each selected block, one unstained slide was submitted for immunohistochemical analysis using a monoclonal antibody directed against the full‐length WT1 protein (DAKO; clone 6F‐H2; dilution 1:40). For this procedure, 3–4‐µm thick sections were cut from paraffin blocks and heated at 58–60°C for 30–60 minutes, deparaffinized, and then rehydrated. Antigen retrieval was performed using DAKO Target Retrieval Solution (pH 9.0 EDTA) for 10 minutes using a pressure cooker. Slides were then stained using the EnVision + Dual Link System™ (Dako North America Inc., Carpinteria, CA, USA). This procedure involved incubation of tissue sections for 5–10 minutes with the Dual Endogenous Enzyme Block (DakoCytomation), incubation with the primary antibody for 60 minutes and incubation with the labeled polymer for 30 minutes, followed by a 5‐minute incubation with a 3‐3′ diaminobenzidine substrate‐chromogen. Slides were then washed and counterstained with hematoxylin. A second unstained slide was submitted for use as a negative control. Positive control tissue, consisting of FFPE sections of Wilms' tumor, was included with each separate run. When present and visible within a given tissue section, the endothelial layer of small blood vessels served as a positive internal control. Sections previously stained with polyclonal anti‐glial fibrillary acidic protein (GFAP) (Dako; Z0334; 1:4000) as part of the original diagnostic work‐up were also reviewed.

IHC scoring

WT1 expression was assessed semiquantitatively for extent of positivity and expression intensity using the scoring systems outlined by Schittenhelm et al (17). The percent of tumor cell positivity was assessed using the following previously published scale: 0 (0%), 1 (<1%), 2 (2%–19%), 3 (20%–50%) and 4 (>50%). Expression intensity was assessed using a second scale: 1+ (weak = blush only), 2+ (moderate) and 3+ (strong = equivalent to vascular positivity). The localization of positive staining was also recorded: n (nucleus), c (cell body), p (cell process) or b (both cell body and cell process).

Electrode monitoring

Nine of the 11 epilepsy specimens were from patients who had undergone prior electrode placement as part of presurgical resection seizure monitoring. Dates of initial electrode implantation, removal and subsequent resection surgery were recorded. The number of days between initial electrode implantation and surgical resection was calculated.

Statistics

For the epilepsy cases with associated electrode monitoring, a correlation coefficient for the relationship between the number of days post‐electrode placement and WT1 expression intensity was calculated.

RESULTS

A summary of WT1 staining results is provided in Table 1. Three of three cases of normal brain tissue (100%) showed WT1 expression within normal‐appearing vascular endothelial cells (Figure 1). No neuronal or glial staining was identified in any of these samples.

Table 1.

WT1 staining in central nervous system tissue samples.

Category	CNS tissue type	No. of cases (n = 100)	WT1 percent positivity scores
			Score 0	Score 1	Score 2	Score 3	Score 4	Pos/total	% Pos
Normal	Normal brain tissue	3	3	0	0	0	0	0/3	0
Neoplastic	Total neoplastic cases	49						49/49	100
	Pilocytic astrocytoma (WHO grade I)	5	0	0	3	2	0	5/5	100
	Oligodendroglioma (WHO grade II)	2	0	0	1	1	0	2/2	100
	Anaplastic astrocytoma (WHO grade III)	2	0	0	0	1	1	2/2	100
	Glioblastoma multiforme (WHO grade IV)	38	0	0	2	7	29	38/38	100
	Giant cell glioblastoma (WHO grade IV)	1	0	0	0	0	1	1/1	100
	Gliosarcoma (WHO grade IV)	1	0	0	0	0	1	1/1	100
Suspicious	Suspicious for glioma	4	0	1	3	0	0	4/4	100
Reactive	Total reactive cases	44						36/44	82
	Hippocampal sclerosis	11	0	5	6	0	0	11/11	100
	Metastatic carcinoma	10	0	1	3	5	1	10/10	100
	Normal pressure hydrocephalus	5	5	0	0	0	0	0/5	0
	Primary tumor—brain interface	8	1	0	2	5	0	7/8	88
	Vascular malformation/Hematoma	7	2	0	2	2	1	5/7	71
	Demyelinating disease/infarct	2	0	0	1	0	1	2/2	100
	Abscess	1	0	0	0	0	1	1/1	100

CNS = central nervous system.

Figure 1.

WT1 staining of normal cortex (×200).

Among glioma cases, 49 (100%) showed WT1 expression, with percent positivity scores ranging from 2 to 4 (median 4) and expression intensity ranging from 2+ to 3+ (median 3+) (Figure 2). Zero cases (0%) showed isolated nuclear staining, 0 cases (0%) showed isolated cell body staining, 2 cases (4.1%) showed isolated cell process staining, and 47 cases (95.9%) showed staining of both the cell body and cell processes.

Figure 2.

WT1 staining of sampled gliomas with positive vessels as internal control. A. Pilocytic astrocytoma showing numerous positive proliferating vessels and weak staining of oligodendroglial‐like cells (×200). Stronger staining of piloid areas seen in inset (2–3+) (×400). B. Oligodendroglioma with mini‐gemistocytes showing cytoplasmic staining (3+) (×400). C. Anaplastic astrocytoma showing strong cell body and cytoplasmic process staining (3+) (×400). D. Glioblastoma showing strong cell body and cytoplasmic process staining (3+) (×400).

Among cases suspicious for low‐grade gliomas, all four (100%) showed WT1 expression, with percent positivity scores ranging from 1 to 2 (median 2) and expression intensity ranging from 2+ to 3+ (median 3+). Zero cases (0%) showed isolated nuclear staining, zero cases (0%) showed isolated cell body staining, one case (25%) showed isolated cell process staining, and three cases (75%) showed staining of both the cell body and cell processes.

Among cases of reactive gliosis, 36 of 44 cases (81.8%) showed WT1 expression, with percent positivity scores ranging from 0 to 4 (median 2) and expression intensity ranging from 0+ to 3+ (median 2+) (Figure 3). Two cases (4.5%) showed isolated nuclear staining, 0 cases (0%) showed isolated cell body staining, 8 cases (18.2%) showed isolated cell process staining, and 32 cases (72.7%) showed staining of both the cell body and cell processes.

Figure 3.

IHC staining of sampled reactive lesions. A. Meningioma with tumor‐brain interface (H&E; ×100). B. GFAP highlighting reactive gliosis at tumor interface (×200). C. WT1 staining of same area showing reactive astrocytes with positive cytoplasmic processes (2+) (×400). D. Cavernous angioma and adjacent cortex (H&E; ×40). E. GFAP showing reactive astrocytes in cortex adjacent to angioma (×200). F. WT1 staining of same area showing delicate fibrillary process positivity in cortex adjacent to angioma (2+) (×200). G. Reactive astrocytes adjacent to metastatic carcinoma (not shown) after radiation treatment (H&E; ×400). H. GFAP staining of the same case showing thick processes of a large reactive astrocyte (×400). I. WT1 staining showing similar positivity in thick astrocyte processes (×400). IHC = immunohistochemistry; H&E = hematoxylin and eosin; GFAP = glial fibrillary acidic protein; IHC = immunohistochemistry.

Nine of the 11 epilepsy cases had presurgical electrode monitoring. The duration of time between initial electrode placement and removal ranged from 7 to 72 days (mean 15.9 days). The time between electrode placement and surgical resection ranged from 7 to 217 days (mean 72.1 days). All nine cases (100%) with electrode monitoring showed WT1 expression—the percent positivity scores ranged from 1 to 2 (median 2) and the expression intensity ranged from 1+ to 3+ (median 2+) (Table 2). In some cases, areas of chronic gliosis confirmed by GFAP staining lacked WT1 expression, while areas adjacent to electrode‐related lesions showed variable WT1 staining intensity (Figure 4). There was a negative correlation between the expression intensity and the number of days after initial electrode placement (R = −0.88). This finding is reflected in the WT1 staining of tissue resected after varying periods of electrode placement (Figure 5).

Table 2.

WT1 staining in post‐electrode monitoring specimens.

Case	Pathologic diagnosis	Tissue	No. days post electrode placement	Positivity score	Expression intensity
1	Cortical dysplasia	Occipital lobe	57	1	3
2	Hippocampal sclerosis	Hippocampus	7	2	3
3	Chronic gliosis	Parietal lobe	7	2	3
4	Hippocampal sclerosis	Hippocampus	72	2	2
5	Hippocampal sclerosis	Hippocampus	100	1	2
6	Hippocampal sclerosis	Hippocampus	56	1	2
7	Hippocampal sclerosis	Hippocampus	63	1	2
8	Hippocampal sclerosis	Hippocampus	70	2	2
9	Hippocampal sclerosis	Temporal lobe	217	2	1

Figure 4.

IHC staining of sampled tissue from epilepsy specimens. A. Temporal cortex with reactive gliosis (H&E, ×100). B. GFAP staining of a single reactive cortical astrocyte (×400). C. WT1 staining of same area as in B showing cytoplasmic positivity within reactive astrocytes and positive vascular staining (3+) (×400). D. Dentate gyrus from patient with hippocampal sclerosis (H&E; ×100). E. GFAP staining highlights the extent of reactive gliosis (×100). F. WT1 staining of the same area as in E showing negative staining of reactive astrocytes (0+; positive internal control vascular staining) (×100). G. Hippocampus (same as D) showing depth electrode lesion in lower left of field (arrow) (H&E; ×20). H. GFAP staining of extensive reactive gliosis surrounding electrode lesion (×100). I. WT1 staining of same area as in H showing weak to moderate expression intensity in reactive astrocytes (1–2+) (×100). IHC =  immunohistochemistry; H&E = hematoxylin and eosin; GFAP = glial fibrillary acidic protein; IHC = immunohistochemistry.

Figure 5.

IHC staining of selected epilepsy specimens after electrode monitoring. A. Hippocampal tissue from Case 2 (Table 2) resected 7 days after electrode placement (GFAP; ×400). B. WT1 staining of same area in A showing 3+ expression intensity (×400). C. Hippocampal tissue from Case 5 (Table 2) resected 100 days after electrode placement (GFAP; ×400). D. WT1 staining of same area in C showing 2+ expression intensity (×400). E. Temporal lobe tissue from Case 9 (Table 2) resected 217 days after electrode placement (GFAP; ×400). F. WT1 staining of same area in E showing 1+ expression intensity (×400). GFAP = glial fibrillary acidic protein; IHC = immunohistochemistry.

DISCUSSION

It has been well documented that when previously resting astrocytes become “activated” in response to a variety of injurious stimuli, there is cellular hypertrophy associated with the up‐regulation of a wide range of intermediate filaments, including GFAP and vimentin (14). Such activated astrocyte populations may arise in response to infectious or inflammatory diseases, infarction and metastatic tumors, among other causes. The degree of cytomegaly and morphologic change may closely mimic malignancy in some cases, making interpretation of biopsy material from such “reactive” lesions quite challenging.

Various antigen expression patterns observed using IHC have been proposed to distinguish between reactive astrocytosis and infiltrating glioma. The presence of strong p53 labeling within an astrocyte population, for example, often supports the interpretation of glioma, just as its positivity in other body sites often supports a diagnosis of malignancy (1). It is also true, however, that a significant number of easily diagnosable gliomas lack p53 expression. Furthermore, as evidence of p53 expression in reactive astrocyte populations has been clearly demonstrated (9), the interpretation of such staining requires caution. Likewise, the expression pattern of the proliferation marker MIB‐1/Ki‐67 also shows some overlap between reactive and neoplastic cases. While typically negative in reactive astrocyte populations, low‐level MIB‐1 expression has been reported, and some low‐grade gliomas may only show minimal MIB‐1/Ki‐67 expression (15). More promising results using IHC have recently been described with antibodies specific for mutant IDH1—tumor cells of diffusely infiltrating astrocytomas are positive while reactive astrocytes are negative (2).

The WT1 protein is a product of the Wilms' tumor suppressor gene (chromosome 11p13), and it is well recognized that mutations or allelic losses of the gene result in nephroblastoma, a common pediatric solid malignancy of the kidney. In addition to playing a role in cell proliferation and differentiation, the WT1 protein may also have oncogenic properties through interaction with other transcription factors (13). WT1 overexpression has been reported in a number of solid tumors, including ovarian carcinoma, sex‐cord stromal tumors, melanoma and mesothelioma (11). The presence of WT1 protein expression has also been reported in various central nervous system gliomas 6, 13.

We were intrigued, then, by the careful and thoroughly documented observations of Schittenhelm et al who reported that WT1 expression is restricted to neoplastic astrocyte populations (17). As in their study, we found that WT1 appears restricted to vascular endothelial cells in normal human brain tissue, and we also confirmed their observation that WT1 shows strong expression in most astrocytic tumors. However, we did observe at least focal WT1 expression in most reactive astrocyte populations, and at least moderate (2+) expression in 68% of cases. Except for biopsies from patients with normal pressure hydrocephalus, all other categories of reactive lesions included in our study showed at least focal WT1 staining. WT1 staining was present both in the cell body and within cell processes, but not within the nucleus. While many of the reactive astrocytes showed more delicate staining of processes with WT1 compared with GFAP, this pattern was not constant. For example, one case of gliosis associated with a previously irradiated metastatic carcinoma showed prominent WT1 staining of the thickened processes and enlarged cell bodies of markedly reactive astrocytes (Figure 4G–I).

Historically, only nuclear expression of WT1 has been interpreted as representing bona fide positive staining—a fact stemming from the confirmed role of the WT1 protein as a transcription factor (10). While cytoplasmic staining has been regarded as nonspecific, some studies have provided evidence that at least one of the WT1 protein isoforms shuttles between the nucleus and cytoplasm in order to possibly influence transcriptional regulation and RNA metabolism (12). Nakatsuka et al have shown predominantly intracytoplasmic expression of WT1 in lung cancer cells by Western blot analysis (11). Thus, while the biological significance of WT1 expression within the cytoplasmic cell body or cell processes alone remains unclear, its presence likely reflects a functional role within the cytoplasm rather than as nonspecific background staining.

The reason for the discrepancy in WT1 expression using IHC most likely stems from differences related to methods of tissue processing and immunohistochemical staining. While the WT1 antibody clone was the same for both studies (Dako 6F‐H2), specimen fixation, antibody dilution and IHC staining platforms differed. First, the Schittenhelm study reportedly used specimens fixed with 4% formalin, while specimens in our study underwent fixation using 10% zinc formalin at all stages of processing. It has been widely reported that zinc formalin solutions provide enhanced immunoreactivity in paraffin‐embedded tissue compared with neutral buffered formalin‐fixed tissue for a broad range of antigens 3, 4, 7, 18. Second, the Schittenhelm study used the antibody at a dilution of 1:50, while we utilized a 1:40 dilution. We did perform staining of a number of cases at a 1:50 dilution, however, and we could not detect a difference in staining intensity. Third, their IHC analysis was performed using the Benchmark^® indirect avidin‐biotin‐based platform (Ventana Medical Systems, Strasbourg, France). IHC analysis in our study was performed using the DAKO EnVision+ system™ (Dako), which is a non‐avidin‐biotin‐based detection method shown to have equal or increased sensitivity compared with indirect avidin‐biotin systems (16). It is therefore plausible that differences in immunohistochemical staining methodology are responsible for these sharply divergent results.

Finally, during our analysis of reactive lesions, we observed differences in WT1 staining intensity among the various reactive astrocyte populations. The epilepsy specimens with associated electrode monitoring provided a source of intentionally produced reactive brain lesions for which precise ages could be calculated. Although the sample size was small, our findings suggest the possibility that there is an inverse relationship between WT1 expression intensity and the age of the reactive lesion. If this conclusion is valid, the underlying mechanism remains unclear.

In summary, the expression of the WT1 transcription factor, which is limited to the vascular endothelium in normal human brain tissue, is seen in both reactive and neoplastic astrocyte populations, contrary to an earlier report. This discrepancy is likely the combined result of differences in tissue fixation and immunohistochemical methodology. Thus, our findings indicate that WT1 expression alone is not a reliable feature to distinguish reactive from neoplastic astrocytes. One interesting finding is that WT1 expression may be related to the timing and duration of astrocyte injury. Future studies will be required to answer this question.