Colloid Cysts of the Third Ventricle:
lmmunohistochemical Evidence for Nonneuroepithelial Differentiation
TAKASHIRA TSUCHIDA, MD, RALPH H. HRUBAN, MD, BENJAMIN S. CARSON, MD, AND PETER C. PHILLIPS, MD
The histogenesis of colloid cysts (CCs) of the third ventricle has been a subject of controversy. We examined, using immunohisto-chemical techniques, four CCs for the presence of cytokeratins (CKs), glutathione S-transferase isoensymes (GST-z, GST-p), and glial fibrillary acidic protein. Antibodies to both low molecular weight CKs (anti-CKS) and to a mixture of CKs (AEl/AES) were used. For comparison, normal fetal and adult choroid plexus, epen-dyma, and nasal mucosa were also examined. The epithelium lining all four CCs showed positive immunostaining for the CKs and GST- z but not for GST-p or glial fibrillary acidic protein. Fetal and adult nasal mucosa showed a pattern of immunohistochemical staining almost identical to that of CCs. In contrast, fetal and adult choroid plexus tissue showed positive immunostaining for GST-r and low molecular weight CKs but not for the CK mixture (AEl/AES). Fetal and adult ependyma were negative for both CKs and GST-r. These results suggest that CCs differentiate along nonneural lines distinct from the neuroepithelial differentiation of the choroid plexus and ependyma. HUM PATHOL 23:811-816. Copyright 0 1992 by W.B. Saunders Company
The histogenesis of colloid cysts (CCs) of the third ventricle has been the subject of considerable contro-versy. Competing hypotheses may be considered in two groups: those suggesting a neuroepithelial origin, in-cluding origin from the choroid plexus,’ ependyma,‘,’ and the embryonic paraphysis,* and those suggesting a nonneural endodermal origin.3-5 Support for each of these hypotheses has been based on various morphologic features observed by light and electron microscopy. The hypothesis that CCs arise from the choroid plexus has been based on similarities in the histologic appearances of their linings and on the observations that choroid plexus tissue can be seen arising from the walls of some CCs.’ The possibility that primitive ependymal tissue can give rise to CCs also has been based on the appear- ance of the epithelium as well as the presence of cilia in both tissues.’ In contrast, electron microscopic studies have suggested that the ultrastructural features of CCs are distinct from those of the choroid plexus and epen- dyma,3 and instead closely resemble those of the epi-thelium of the trachea’ and sphenoid sinus.4,5
From the Departments of Neurosurgery, Pathology, Neurology, Oncology, and Pediatrics, The Johns Hopkins Hospital, Baltimore, MD. Accepted for publication September 19, 1991.
Key words: histogenesis, colloid cyst, third ventricle, cytokeratin, glutathione S-transferase, immunohistochemistry.
Address correspondence and reprint requests to Ralph H. Hru- ban, MD, Department of Pathology, The Johns Hopkins Hospital, 600 N Wolfe St, Baltimore, MD 21205.
Copyright 0 1992 by W.B. Saunders Company
0046-8177/92/2307-0016$5.00/O
To determine whether immunohistochemical staining characteristics could be used to assist in the identification of the direction of differentiation of CCs, we compared four cases of CCs of the third ventricle to normal human adult and fetal choroid plexus, epen-dyma, and nasal epithelium. Immunohistochemical stains for broad-spectrum and low molecular weight cy- tokeratins (CKs), glutathione S-transferase-r (GST-?r), glutathione S-transferase-p (GST-CL), and glial fibrillary acidic protein (GFAP) were used. Our results suggest that CCs differentiate along nonneural lines similar to the epithelium of the trachea and sphenoid sinus.


MATERIALS AND METHODS
Tissue
Four patients underwent surgical resection of CCs of the third ventricle between 1980 and 1989 at The Johns Hopkins Hospital. Resected tissue was fixed in 10% phosphate-buffered formalin and embedded in paraffin. In two of the four cases, normal choroid plexus tissue was attached to the CC. For pur-poses of comparison, normal choroid plexus and normal ependyma were obtained from tissue resected from patients with intractable epilepsy. Nasal mucosa was obtained from transsphenoidal hypophysectomies. Human choroid plexus, ependyma, and nasal mucosa were obtained from two fetuses aborted at the ninth and 17th week of gestation. Positive con-trol tissues for GST-n, GST-p, and GFAP included human placenta, human liver, and normal human cerebral cortex, respectively. All tissues were sectioned at a thickness of 5 mp.
Antibodies
A monoclonal antibody to broad-spectrum CKs (AEl/ AE3) was obtained from Hybritech Inc (San Diego, CA). This mixture of antibodies contains the anti-CK monoclonal anti-body AEl, which recognizes the 50- and 56.5-kd CKs (CKs 10 and 14),6 and AE3, which recognizes the 58-, 65.5-, and 68-kd CKs (CKs 1, 2, and 5).‘j A monoclonal antibody to low molecular weight CK was obtained from ENZO Biochem Inc (New York, NY). This antibody recognizes the 52.5-kd CK (CK 8) (manufacturer’s data). Rabbit anti-cow GFAP antibody was obtained from DAK0 Co (Carpinteria, CA).
Rabbit antiserum to human GST-7r’ and affinity-purified rabbit antibody to human GST-p were obtained from Medlabs (Dublin, Ireland). The specificities of these antibodies were confirmed on immunoblotting following sodium dodecyl sul-fate polyacrylamide gel electrophoresis in which purified GST-s and GST-r are run (MEDLABS, Dublin, Ireland).
lmmunohistochemistry
Deparaffinized sections were pretreated with 0.03% H202 in methanol for 30 minutes at room temperature to block

811
endogenous tissue peroxidase activity.8 Sections for CK stain- ing were also predigested with pronase (Calbiochem, San Diego, CA) 1 mg/mL in 0.05 mol/L tris-buffered saline sup-plemented with 0.23% EDTA for 30 minutes at 37°C. Pronase activity was arrested by immersing the sections in 95% ethyl alcohol for 5 minutes. To block nonspecific antibody binding, sections stained for CST-p, GST-*, and GFAP were incubated with normal goat serum (1:lO) for 30 minutes at room tem- perature. For CK staining, sections were incubated with normal horse serum (1:20) for 30 minutes at room temperature. Sec-tions were then incubated with the following primary anti-bodies at the dilutions indicated: AEl/AEJ (1:2,000 dilution), anti-CK 8 (1:2,000 dilution), GFAP (1:500 dilution), GST-T
(1: 1,000 dilution), or GSTq (1: 100 dilution), all for 12 hours at 4°C. Normal rabbit serum and normal mouse immuno-globulin G (IgG) were used as negative controls. Sections were incubated with the secondary antibody: biotinylated anti-mouse IgG horse immunoglobulin (Vector Laboratories Inc. Burlin-game, CA) for CK or biotinylated anti-rabbit IgG goat Ig (Vec- tor Laboratories Inc) for 30 minutes at room temperature. Sections for CK staining were incubated with peroxidase-con-jugated streptavidin (1: 1,000 dilution) (Jackson Immuno Re-search Laboratories Inc, West Grove, PA); all other sections were incubated with avidin-biotin-peroxidase complex (Vector Laboratories Inc) for 30 minutes at room temperature.g Sec-tions were immersed in aminoethyl carbazole (Biomeda Corp, Foster City, CA) for 10 minutes or in diaminobenzidine so-lution (Sigma, St Louis, MO) for 5 minutes and then coun-terstained with hematoxylin.


RESULTS
Colloid Cysts
Hematoxylin-eosin-stained sections of the four CCs of the third ventricle demonstrated that the epithelial lining of the CCs presented here was fundamentally the same epithelial lining (ie, ciliated and nonciliated cu-boidal to columnar epithelium) as is seen in the nasal mucosa. The percentage of ciliated cells varied among cases and in two of the four cases, normal colloid plexus tissue was attached to the CC (Fig 1).
Immunohistochemical staining of the four CCs re- vealed intense staining for broad-spectrum CKs in all four cases (Fig 1). The staining was localized to the cy- toplasm of ciliated and nonciliated columnar to cuboidal epithelia. A similar staining pattern was observed for the low molecular weight CK (Fig 2), but the intensity of staining was less. All four cases also stained for GST- 7~ (Fig 3). Patchy positive staining was observed in both the cytoplasm and nucleus of ciliated and nonciliated columnar to cuboidal epithelia. In contrast, positive staining for GST-p and GFAP was not observed.
Normal Tissues
Fetal and adult choroid plexus tissues were also examined. In these tissues staining for broad-spectrum CKs was not identified, while positive staining for low molecular weight CK was observed in the cytoplasm of both fetal and adult choroidal epithelia. Positive staining for GST-?r was found mainly in the cytoplasm of fetal and adult choroidal epithelia, but occasional nuclear staining was identified in adult choroidal epithelia. Pos-itive staining for GST-p or GFAP was not detected in these tissues.
Fetal and adult ependyma were also examined, and positive staining was not observed with any of the an-tibodies used. Stains for broad-spectrum and low mo- lecular weight CKs, for the GSTs, and for GFAP were all negative.
Fetal and adult nasal mucosa stained positively fol both broad-spectrum and low molecular weight CKs. Positive staining for GST-x was observed in both the cytoplasm and nucleus of ciliated columnar epithelia of the mucosa. Positive staining for GST-p was not observed and GFAP staining was not performed. A summary of these results is listed in Table 1.


DISCUSSION
The histogenesis of CCs of the third ventricle is controversial. Paraphysis,’ neuroepithelium,’ and en-doderm3 all have been reported as possible sites of origin based on the morphologic similarity of the cyst epithelial lining to these structures. This confusion over the “cell of origin” of CCs is mirrored in the nomenclature ap-plied to these lesions. For example, CCs have been re-ferred to as ependymal cysts, choroidal epithelial cysts, neuroepithelial cysts, and epithelial cysts.‘,10-12
The majority of the initial investigations into the histogenesis of CCs were based on detailed light micro- scopic observations. In 1955, Kappers’ suggested that most CCs of the third ventricle were derived from either the diencephalic vesicle (ie, ependyma) or the paraphysis, a rudimentary organ in the human fetus that disappears by 3.5 months of gestation. Ten years later, however, Shuangshoti et al’ suggested that there was no mor-phologic difference between diencephalic vesicle and paraphysis, and proposed that CCs of the third ventricle are derived from the neuroepithelium common to both the paraphysis and the diencephalic vesicle. Based on their observations, Shuangshoti et al suggested that the term “colloid cyst” be replaced by the term “neuro-epithelial cyst.“’
The introduction of electron microscopy has added to the understanding of the direction of differentiation of CCs. In 1974, Hirano and Ghatak examined three CCs of the third ventricle and noted that the fine struc- ture of the epithelial cells lining these cysts differed from that of normal choroidal epithelium and ependyma.3 The cysts they examined were lined by two types of cells, a lighter ciliated cell and a darker nonciliated cell. These cells rested on a basement membrane, and the noncili- ated cells had microvilli with a constant diameter and coated substance on their luminal surfaces. Hirano and Ghatak noted the ultrastructural similarities between the epithelium lining CCs of the third ventricle and the tra- cheal epithelium and, based on these observations, speculated that CCs of the third ventricle might arise from midline endodermal diverticula during embryo-genesis.” In support of this idea, epithelial cysts in the fourth ventricle ’ and spinal cordi with morphologic appearances similar to those of CCs of the third ventricle have been reported, while ependyma usually lacks a basement membrane and choroidal epithelia have club- shaped microvillL5
The purpose of this study was to examine, using immunohistochemical techniques, the direction of dif-

FIGURE 1. (Top) lmmunohistochemical stain for CK (AEI/AE3) demonstrating that the epithelium of this CC (arrow) is strongly pos-itive for CK while the normal choroid plexus (arrowhead) attached to the cyst wall is negative. (CK AE 1/AE3; magnification x50.) (Bottom) High-power view of the top panel showing intense staining for CK in the colum- nar to cuboidal epithelium of the CC. (CK AEI/AW: magnification x400.)
ferentiation of CCs of the third ventricle. Antibodies co CKs, to GSTs, and to GFAP were used. The anti-CK antibodies used in this study included an antibody to low molecular weight CKs and a mixture of antibodies against CKs (AEl/AE3).‘j The mixture of the anti-CK antibodies AEl and AE36 recognizes the CKs 1, 2, 5, 10, and 14, which are widely distributed in various kinds of epithelium.15 It is interesting to note that CKs (AEl/ AE3) have not been detected in the choroidal epithelia or in the ependyma, although it has been reported that low molecular weight CKs (CK 8, 18, and 19) are present in human choroidal epithelia as well as in CCs of the third ventricle.‘6*‘7 We found the epithelial lining of CCs of the third ventricle to be strongly positive for the mix- ture of CKs (AEl/AE3). Although the intensity of stain- ing was less, a similar pattern of staining was observed when antibodies to the low molecular weight CKs were used. This positive staining for low molecular weight CKs suggests that the epithelium lining CCs of the third ventricle is distinct from choroidal epithelium.
The second v ou p of antibodies used in this study was to the GSTs. GIutathione S-transferases are ubiq-uitous intracellular enzymes that participate in the me-tabolism and detoxification of a wide range of electro- philic compounds by conjugation with glutathione.’ The GST isoenzymes are grouped by their isoelectric points into three classes: basic, GST-cu-t; near-neutral, GST-pu; and acidic, GST-*. We chose to examine the patterns of expression of GST-?r and GST-p in this study because they are potentially useful markers of differentiation in

FIQURE 2. The epithelium of both ccs (top) and choroid plexus (bottom) stalnel d positively with antibodies against the low molecular weight CKs. (CK 8. Magnifications: tOD, X400; bottom, x200.)

the brain. Glutathione S-transferase-l* and GST-x have nohistologic results indicate that the epithelium lining been identified in normal human brain,‘8-20 GST-?r has CCs of the third ventricle is similar to nasal mucosa and been found in normal choroid plexus, and the expres- unlike ependyma and choroidal epithelia (ie, neuroep- sion of GST-r has been found to increase in various ithelium). These immunohistochemical results are not premalignant21~22 and malignant neoplasms.21923-27 We entirely surprising. Kondziolka and Bilbao” reported found that GST-7r was not expressed in human epen-that CCs of the third ventricle are distinct from choroid dyma, although it was demonstrated in choroidal epi-plexus based on the staining of CCs with pan-epithelial thelia and nasal mucosa. The epithelium lining CCs of and pan-CK antibodies and the absence of staining with the third ventricle stained positively for GST-?r. these same antibodies in the choroid plexus. Our finding
The third group of antibodies used in this study that both CCs and choroid plexus stain positively for were against GFAP. Glial fibrillary acidic protein was the low molecular weight CKs, while only CCs stain for not expressed in the epithelial lining of the four CCs in the broad-spectrum CKs, suggests that it is the high mo- our cases, although ependyma should express GFAP at lecular weight CKs that distinguish CCs from the choroid certain developmental stages.*‘“’ In toto these immu-plexus.
FIGURE 3. lmmunohistochemical stain for GST-X. (Top) Note that the epithelium of both the CC (arrow) and the attached choroid plexus (arrowhead) stain positively for GST-
r.
(GST-r; magnification x35.) (Bottom) Higher-power view of the top panel dem-onstrating that the columnar to cuboidal epithelium of the CC stains positively for GST-

X.
(GST-T; magnification x200.)


Of interest, the fine structure of the epithelial lining of CC also is very similar to that of symptomatic Rathke’s cleft cysts and to the mucosa of the sphenoid sinus.4v5 The epithelial lining of the CCs presented here also showed immunohistologic as well as morphologic simi-larities to nasal mucosa. For example, intracellular mu-tin has been found in a substantial number of CCs but is usually not present in choroid plexus epithelium.” It is interesting to note that Rathke’s cleft and nasal mu-cosa are considered to be derived from similar origin (ie, stomodeum).32s3
While the results of the immunohistochemical studies presented here suggest that CCs of the third
ventricle differentiate along lines similar to those of the nasal mucosa, two of the four cysts we examined were closely associated with histologically normal choroid plexus tissue. This finding emphasizes the important distinction between differentiation and site of origin of tumors. 34 Although there is a general
correspondence  between  morphologic  characteris-
tics  and  organ  of  origin,  striking  exceptions  have
been  reported.  For  example,  hepatocellular  carci-

nomas of the pancreas and stomach have been re-ported.34
Our studies help define the direction of differen- tiation of CCs. This direction of differentiation is along
TABLE 1. lmmunohistochemical Results for Colloid
Cysts, Neuroepithelium, and Nasal Mucosa

Broad Spectrum LMW GST GST
Age (yr)/Sex CK CK P p GFAP
Colloid cysts 30/M + + + -	-
_

60/M + + + -_
55/M + + + -
62/F + + + --
Ependyma

8/M 	-----
_ -	-
18/F _ -_ _
35/M 9wk* -----17wk* ----_ _
Choroid + + -	-
2/F _ _
plexus 11/M + + -13/F -+ + --_
38/M 	-+ + -_
39/M + + -	-_
55/M 	-+ + -
9 wk* 	-+ + --_
17 wk* f + -	-

Nasal 55/F + + 	+ -NP
mucosa 9 wk* + + + -NP
17 wk* 	+ + + -NP

Abbreviations: CK, cytokeratin; LMW, low molecular weight; GST, glutathione S-transferase; GFAP, glial fibrillary acidic protein; NP, not performed.
* Fetal age in gestational weeks.

nonneuroepithelial lines, distinct from those of the ependyma and choroid plexus.
Acknowledgment. The authors thank Wes Gage, George

L. Pettis, Amy Pielert, and Claire E. Hruban for their assis- tance.

REFERENCES
1, Shuangshoti S, Roberts MP, Netsky MG. Neuroepithelial (col-loid) cysts: Patbogenesis and relation to choroid plexus and ependyma. Arch Path01 80:214-224, 1965
2.
Kappers JA: The development of the paraphysis cerebri in man with comments on its relationship to the intercolumnar tubercle and its significance for the origin of cystic tumors in the third ventricle. J Comp Neurol 102:425-498, 1955

3.
Hirano A, Ghatak NR: The fine structure of colloid cysts of the third ventricle. J Neuropathol Exp Neurol 33:333-341, 1974

4.
Matsushima T, Fukui M, Kitamura K, et al: Mixed colloid cyst-xanthogranuloma of the third ventricle. Surg Neurol24:457-462,1985

5.
Matsushima T, Fukui M, Fujii K, et al: Epithelial cells in symp- tomatic Rathke’s cleft cysts: A light and electron-microscopic study. Surg Neurol 30:197-203, 1988

6.
Woodcock-Mitchell J, Eichner R, Nelson WG, et al: Immu-nolocalization of keratin polypeptides in human epidermis using monoclonal antibodies. J Cell Biol 95:580-588, 1982

7.
Mannervik B: The isoenzymes of glutathione transferase. Adv Enzymol 57:357-417, 1985

8.
StreeIkerk JG: Inhibition of erythrocyte pseudoperoxidase activity by treatment with hydrogen peroxide following methanol. J Histochem Cytochem 20:829-831, 1972

9.
Hsu S-M, Raine L, Fanger H: Use of avidin-biotin-peroxidase complex (ABC) in immunoperoxidase techniques: A comparison be-tween ABC and unlabeled antibody (PAP) procedures. J Histochem Cytochem 29:577-580, 1981


10.
Hoenig EM, Ghatak NR, Hirano A, et al: Multiloculated cystic tumor of the choroid plexus of the fourth ventricle: Case report. J Neurosurg 27:574-579, 1967

11.
Andreussi L, Cama A, Cozzutto C, et al: Cyst of the choroid
plexus of the left lateral ventricle. Surg Neurol 12:53-57, 1979


12.
Inoue T, Kuromatsu C, Iwata Y, et al: Symptomatic choroidal
epithelial cyst in the fourth ventricle. Surg Neurol 24:57-62, 1985


13.
Parkinson D, Childe AE: Colloid cyst of the fourth ventricle: Report of a case of two colloid cysts of the fourth ventricle. J Neurosurg 9:404-409, 1952

14.
Hirano A, Ghatak NR, Wisoff HS, et al: An epithelial cyst of the spinal cord. An electron microscopic study. Acta Neuropathol 18: 214-223, 1971

15.
Moll R, Franke WW, Schiller DL, et al: The catalog of human cytokeratins: Patterns of expression in normal epithelia, tumors and cultured cells. Cell 31:l l-24, 1982

16.
Miettinen M, Clark R, Virtanen I: Intermediate filament proteins in choroid plexus and ependyma and their tumors. Am J Path01 123:231-240, 1986

17.
Kouno M, Kumanishi T, Washiyama K, et al: An immuno-histochemical study of cytokeratin and glial fibrillary acidic protein in choroid plexus papilloma. Acta Neuropathol 75:317-320, 1988

18.
Polidoro G, Di Ilio C, Sacchatta G, et al: Isoelectric focusing of brain cortex GSH S-transferase activity in mammals: Evidence that polymorphism is absent in man. Int J Biochem 16:741-746, 1984

19.
Theodore C, Singh SV, Hong TD, et al: Glutathione S-trans-ferases of human brain: Evidence for two immunologically distinct types of 26500-Mr subunits. Biochem J 225:375-382, 1985

20.
Corrigall A, Kirsh R: Glutathione S-transferase distribution and concentration in human organs. Biochem Int 16:443-448, 1988


2 1. Kodate C, Fukushi A, Narita T, et al: Human placental form of glutathione S-transferase (GST Pi) as a new immunohistochemical marker for human colonic carcinoma. Jpn J Cancer Res 77:226-229, 1986
22.
Shiratori Y, Soma Y, Maruyama H, et al: Immunohistochem- ical detection of the placental form of glutathione S-transferase in dysplastic and neoplastic human uterine cervix lesions. Cancer Res 47:6806-6809, 1987

23.
Soma Y, Satoh K, Sato K: Purification and subunit-structural and immunological characterization of five glutathione S-transferases in human liver, and acidic form as a hepatic tumor marker. Biochem Biophys Acta 869:247-258, 1986

24.
Shea TC, Kelly SL, Henner WD: Identification of an acidic form of glutathione transferase present in many human tumors and human tumor cell lines. Cancer Res 48:527-533, 1988

25.
Niitsu Y, Takahashi Y, Saito T, et al: Serum glutathione S transferase Pi as a tumor marker for gastrointestinal malignancies. Cancer 63:317-323, 1989

26.
Campbell JAN, Corrigall AV, Guy A, et al: Immunohistologic localization of alpha, mu, and pi class glutathione S-transferases in human tissues. Cancer 67:1608-1613, 1991

27.
Tsuchida T, Carson B, Hruban RH, et al: Immunohisto-chemical detection of glutathione-s-transferase Pi (GST-n) in brain tumors. Proc Am Assoc Cancer Res 31:354, 1990 (abstr)

28.
Roessmann U, Velasco ME, Sindely SD, et al: Glial fibrillary acidic protein (GFAP) in ependymal cells during development. An immunocytochemical study. Brain Res 200: 13-21. 1980

29.
Mannoji H, Becker LE: Ependymal and choroid plexus tu-mors. Cytokeratin and GFAP expression. Cancer 61:1377-1385, 1988

30.
Duffy PE, Graf L, Huang Y-Y, et al: Glial fibrillary acidic protein in ependymomas and other brain tumors. J Neurol Sci 40: 133-146, 1979

31.
Kondziolka D, Bilbao JM: An immunohistochemical study of neuroepithelial (colloid) cysts. J Neurosurg 71:91-97, 1989

32.
Fager CA, Carter H: Intrasellar epithelial cysts. J Neurosurg 24:77-81, 1966

33.
ValdCs-Dapena MA: The pituitary, in Valdes-Dapena MA (ed): Histology of the Fetus and Newborn. Philadelphia, PA, Saunders, 1979, pp 103-115

34.
Hruban RH, Molina JM, Reddy MN, et al: A neoplasm with pancreatic and hepatocellular differentiation presenting with subcu-taneous fat necrosis. Am J Clin Path01 88:639-645, 1987