9.1 Introduction

The choroid plexus has the highly specific function of producing cerebrospinal fluid (CSF). It is anatomically localized to the parenchymal/ventricular junction in all four ventricles. The choroid plexus is derived from the specialization of ventricular epithelium along certain segments of the neural tube, and there is a common ontogeny between choroid epithelium and cells of glial origin. This can, and does lead to diagnostic confusion in certain cases. Tumors arising from the choroid plexus can display a benign or malignant phenotype, but conversion to a malignant phenotype is a rare event (Chow et al. 1999; Jeibmann et al. 2007). Guerard was the first to describe a choroid plexus tumor in 1833. The first surgical resection was reported by Bielschowsky and Unger in 1906. Thereafter, both Cushing and Dandy reported their experiences with this unusual tumor (Dandy 1922; Davis and Cushing 1925).

9.2 Epidemiology

Choroid plexus papilloma (CPP) and choroid plexus carcinoma (CPC) are uncommon, comprising only 0.5–0.6% of all brain tumors. Although found in all age groups, choroid plexus neoplasms are primarily a tumor of childhood. Laurence in his review of all published cases prior to 1974 reported that 45% presented in the first year of life, while 74% were in the first decade (Laurence 1974). As expected, reviews from pediatric centers report that a higher percentage (1.8–2.9%) of their cases are choroid plexus tumors (Asai et al. 1989; Ellenbogen et al. 1989; Sarkar et al. 1999). In two reviews describing tumors occurring in the first year of life, choroid plexus tumors comprised 14 and 12.8% of all cases (Galassi et al. 1989; Haddad et al. 1991). The majority of the series noted here have not reported any predilection for right or left ventricle, or sex. Laurence did report that 50% of cases reviewed were situated in the lateral ventricles, 37% in the fourth ventricle, 9% in the third ventricle, and the remainder in other locations. Other series have confirmed this geographic distribution. CPCs, while rare, comprise 29–39% of all choroid neoplasms (Ellenbogen et al. 1989; Johnson 1989; St Clair et al. 1991).

9.3 Pathology

9.3.1 Gross Appearance

CPP is frequently described as “cauliflower-like.” Indeed, these tumors are similar to the soft fronds of normal choroid, as found in the ventricles. The shape is roughly globular, with an irregular surface and intervening encapsulated areas. Old hemorrhage is sometimes apparent. Since papillomas are benign, they tend to expand the ventricle rather than invade the adjacent brain. Nevertheless, the proximity of these tumors to deep-seated structures such as the internal cerebral veins and limbic structures can make their removal difficult.

9.3.2 Histopathology

CPP is a WHO grade I tumor, and its microscopic appearance recapitulates the normal choroid plexus. There are many papillae covered with a simple cuboidal or columnar epithelia. The stroma of these fibrovascular structures is composed of connective tissue and small blood vessels. The presence of the connective tissue stroma is notable mainly because it allows one to distinguish between CPP and papillary forms of ependymoma (whose stroma is composed of fibrillary neuroglia). In addition, choroid epithelial cells do not contain cilia or blepharoplasts as do ependymal cells. Mitotic figures are rare.

Villous hypertrophy of the choroid plexus is a poorly defined entity. Characteristically, the choroid plexus of both lateral ventricles is enlarged and is associated with hydrocephalus from birth. Russell and Rubinstein comment that the hydrocepha-lus is related to hyperactivity of the choroid while the cytological appearance of the tissue is normal (Bigner et al. 1998). Other authors have used vil-lous hypertrophy synonymously with bilateral CPP, but this is not accurate in the strictest sense if his-tologic evidence of neoplastic growth is not present, and expansion of the choroid plexus occurs diffusely (Hirano et al. 1994).

CPCs are WHO grade III tumors and are diagnosed on the basis of their microscopic appearance (Gopal et al. 2008). Two major features accompany malignancy. First is the presence of brain invasion by the tumor. This usually involves transgression of the ependymal lining and extension into the paraventricular parenchyma. Second, cytological criteria of malignancy — nuclear atypia, increased nuclear to cytoplasmic ratio, prominent mitotic figures, and necrosis — are present in association with a loss of normal papillary architecture. Rarely, if a tumor demonstrates some atypical features without evidence of invasion, it can be designated as an atypical papilloma. The epithelial nature of the frank malignancy can create confusion, since other tumors such as metastatic adenocarcinoma, papillary meningioma, and atypical teratoid/rhabdoid tumors (ATRT) can be histologically similar. If the tumor arises in a young patient, then chances of the tumor being metastatic are extremely low. Electron microscopy can reveal details such as cilia, which are normally not present in choroid plexus tumors. Grossly, these tumors tend to be softer and more friable than papillomas. While carcinomas rarely metastasize from the intracranial or intraspinal compartment, they can disseminate throughout the CSF pathways (McComb and Burger 1983).

An intermediate entity, the atypical CPP, is identi-fied as a WHO grade II neoplasm but the diagnostic criteria are poorly defined (Paulus and Brandner 2007). The number of mitotic figures in a high-power field, or two other cytological features such as increased cellularity, nuclear pleomorphism, and/or necrosis, have been proposed as characteristics of atypical papillomas but this requires validation (Jeib-mann et al. 2007). It is likely, however, that a biological spectrum exists for tumors identified as papillomas and the clinician should be alert to unusual pathologic features that would prompt closer surveillance imaging in the postoperative period.

9.3.3 Immunohistochemistry

Only a few immunohistochemical stains have been found to be helpful. The calcium-binding protein S-100 is positive in the vast majority of choroid tumors (Paulus and Janisch 1990; Ho et al. 1991). This is of limited value since glial tissues and normal choroid express S-100 in a parallel fashion with glial fibrillary acid protein (GFAP). Other markers such as vimentin, GFAP, and cytokeratins can be positive but they also lack specificity (Mannoji and Becker 1988; Cruz-Sanchez et al. 1989). Pre-albumin, or transthy-retin (TTR), was initially believed to be a specific marker, but another report noted that 20% of chor-oid tumors were TTR- negative (Herbert et al. 1990; Paulus and Janisch 1990). These investigators did find that prognostic information could be gleaned from immunohistochemical data. A poor prognosis was found in those tumors with less than 50% of the cells in a given tumor heavily stained for S-100. In addition, absence of TTR-positive cells correlated with a poor prognosis. Cellular proliferation, as measured by Ki67/MIB-1 labeling, is low with papillomas and sig-nificantly higher for carcinomas (Vajtai et al. 1996).

Using a microarray approach, Hasselblatt et al. identified a number of genes that appeared to be over-expressed in choroid plexus tumors (Hasselblatt et al. 2006). Two in particular, Kir7.1 (a potassium channel gene) and stanniocalcin-1, demonstrated high speci-ficity and were proposed as markers for these tumors, but require confirmation by other groups. Finally, Judkins et al. noted that the BAF47 clone of the INI1 gene product was expressed in the majority of CPCs, but not in atypical teratoid/rhabdoid tumors, so this marker may be useful to distinguish between these tumor types (Judkins et al. 2005).

9.3.4 Genetics

The cause of choroid plexus tumors is unknown. One report has mentioned two cases occurring in one family, but a hereditary basis has not been observed for most cases (Zwetsloot et al. 1991). There is some evidence linking SV40, a primate DNA virus, with choroid plexus tumor etiology. Large T antigen, the major regulator of late viral gene products of the SV40 virus, when expressed in mice induces the formation of choroid plexus neoplasms (Brinster et al. 1984). The large T antigen is expressed only in the choroid plexus and appears to interact with the product of the p53 gene (Marks et al. 1989). Using PCR, SV40 DNA sequences were demonstrated in 50% of choroid plexus tumors and the majority of ependymomas (Bergsagel et al. 1992). Active T antigen and p53 complexes have also been demonstrated in brain tumors (Zhen et al. 1999). Positive nuclear staining for the p53 tumor suppressor gene was identified in 10 of 11 CPCs, but in only 1 of 12 CPPs (Carlotti et al. 2002). A p53 mutation in this setting leads to a loss of normal gene function but an increased half-life of the protein. Germline mutations in p53 can also lead to the development of CPCs (Krutilkova et al. 2005).

Experiments in mice showed that the expression of transgenes of the viral oncoproteins E6 and E7 from human papilloma virus produced tumors in 71% of offspring, and 26% of the tumors were chor-oid plexus tumors (Arbeit et al. 1993). Finally, mice that overexpress the E2F1 gene in glial cells develop tumors such as medulloblastoma, CPCs, and primitive neuroectodermal tumors (PNETs) at an early age (Olson et al. 2007).

A subset of central PNETs, CPCs, and medullo-blastomas were recently shown to have frequent mutations in the hSNF2/INI1 gene, which encodes for a component of the ATP-dependent chromatin remodeling complex (Sevenet et al. 1999a). The same authors have proposed that constitutional mutations in this gene lead to a greater incidence of renal and extrarenal malignant rhabdoid tumors, CPCs, central PNETs, and medulloblastomas: a complex they have coined the “rhabdoid predisposition syndrome” (Sevenet et al. 1999b). The penetration of the disease is high, with many probands developing malignant tumors before 3 years of age. Some pathologic data also suggest a connection between malignant rhab-doid tumors and CPCs (Wyatt-Ashmead et al. 2001).

A number of chromosomal abnormalities have been identified in both CPP and CPC. Tumors with a gain of 9p and loss of 10q are associated with longer survival (Rickert et al. 2002). Surprisingly, even benign CPP (32 of 34 cases) demonstrated chromosomal aberration. The patterns of aberrations in CPP differ from those observed in CPC.

9.4 Clinical Features

Hydrocephalus is the presenting symptom in the vast majority of patients with choroid plexus tumors. It is caused by both overproduction of CSF and, in certain cases, the obstruction of CSF pathways, although it appears that overproduction is the major factor (Eisenberg et al. 1974). Resolution of hydrocephalus has been reported after complete tumor removal, suggesting that CSF hypersecretion was responsible for ventriculomegaly (Matson and Crofton 1960; Wilkins and Rutledge 1961; Gudeman et al. 1979). Variations are likely to exist since a normal rate of CSF production has been reported in a patient with a papilloma (Sahar et al. 1980).

The most common presentation of choroid plexus neoplasms is related to increased intracranial pressure secondary to obstructive hydrocephalus and/ or CSF overproduction (Laurence 1974; Humphreys et al. 1987; Ellenbogen et al. 1989). Since the majority of cases occur in infants and young children there are characteristic features of raised ICP; Ellenbogen described the usual presenting signs and symptoms (Ellenbogen et al. 1989). The most common symptoms described were nausea/vomiting, irritability, headache, visual difficulty, and seizure. As expected, the most common signs were craniomegaly, papille-dema, and decreased level of consciousness. The duration of symptoms reported in this series varied from 2 months in those patients younger than 2 years of age to 6 months on average in those patients older than 2 years. Although choroid neoplasms are viewed as slow-growing tumors, the presence of stupor or coma as the presenting sign in 25% of children suggests a more acute clinical course in some patients. Rapid decompensation can occur either from massive hydrocephalus or from tumoral hemorrhage. Of 21 patients who had CSF examined, 2 were found to have grossly bloody fluid. Lateralizing signs are found in a minority of patients and are usually related to asymmetrical ventricular dilatation. Hydrocephalus was present in 78% of cases at the Hospital for Sick Children, and in 95% of cases at the Children's Hospital in Boston (Humphreys et al. 1987; Ellenbogen et al. 1989).

9.5 Diagnosis and Neuroimaging

Since most patients present with hydrocephalus and increased intracranial pressure, there is no role for sampling CSF at diagnosis. There is little information that can be gained from CSF sampling, and there are reports of disastrous outcomes in some patients following lumbar puncture (Laurence 1974). No spe-cific laboratory tests are available to diagnose these tumors. Other benign lesions of the choroid plexus such as choroid plexus cysts, villous hyperplasia, and lipomas can usually be distinguished on the basis of their appearance on magnetic resonance (MR) imaging (Naeini et al. 2009).

9.5.1 Computed Tomography

The typical features of CPP are present on a computed tomography (CT) scan. The mass is well-demarcated from the brain, lobulated, and often has punctate calcification. These tumors enhance homogeneously after contrast, reflecting a luxuriant blood supply (Laurence 1974). Since they arise from the choroid plexus, their location is almost always intraventricular. An enlarged choroidal artery leading into the tumor mass can sometimes be seen in postcontrast images. At times, the massive size of these lesions may obscure the site of origin. Some carcinomas display a diffuse border between tumor and normal brain that may reflect areas of brain invasion. On the basis of CT, certain features distinguish a suspected choroid tumor from other possibilities. Cerebellar astrocytomas tend to be less homogeneously staining and often have cystic areas. Medulloblastomas are characterized by a more heterogeneous appearance, although they also stain vividly with contrast and may cause confusion with a fourth ventricle choroid papilloma. Finally, ependymomas arise physically in similar locations but tend to enhance inhomogeneously.

9.5.2 Magnetic Resonance Imaging

Papillomas are isointense to brain on T1-weighted images (Fig. 9.1a). Areas of high signal indicate hemorrhage or necrosis. Following gadolinium administration, the tumor enhances brightly (Fig. 9.1b, c), although this can be patchy in nature, reflecting areas of high flow. T2-weighted images demonstrate an intermediate to high signal intensity with areas of heterogeneous internal signal (Coates et al. 1989). With CPC, the boundary between the tumor and surrounding brain can be indistinct in areas, but this is not a universal finding (Meyers et al. 2004). Brain edema surrounding a CPC is often observed.

Figure 9.1
figure 1

A large choroid plexus papilloma. (a) A sagittal precontrast T1-weighted MRI demonstrates a large lobulated mass within the lateral ventricle with associated enlargement of the ventricle. (b, c) Following contrast, the mass enhances brightly. Note that the papilloma is well demarcated from the ventricular wall

Full size image

MR spectroscopy of CPP and CPC is characterized by a prominent choline peak and absence of N -acetyl aspartate (Horska et al. 2001). Myo-inositol level is also reported to be specifically increased in CPPs (Krieger et al. 2005). As with CT, an enlarged choroidal artery is often noted, espe-cially with larger tumors. The vascularity of these tumors is easily demonstrated with specific perfusion sequences (Fig. 9.2).

Figure 9.2
figure 2

a) An axial postcontrast image shows a large choroid plexus papilloma within the temporal horn of the lateral ventricle. (b) The perfusion sequence results in a “negative” image with increased vascularity depicted as a dark area. The mass is considerably darker than the adjacent brain tissue

Full size image

9.6 Treatment

9.6.1 Preoperative Planning

Since most patients present with symptoms of intrac-ranial hypertension, the order and type of treatment is directed at relieving hydrocephalus, determining the diagnosis, and removing the tumor. Unless the patient is rapidly deteriorating, urgent CSF drainage is not necessary. At the time of surgery, a ventricular drain is placed in order to reduce brain tension and allow sufficient retraction. An external ventricular drain may be left in place after surgery in order to monitor ICP and to determine if shunting is required in the early postoperative period. Matson and others have reported that the successful removal of a tumor obviates the need for shunting. However, it is likely that other factors such as ventricular bleeding, postoperative changes, or meningitis can also render the patient shunt-dependent. Ellenbogen's series noted that 37% of surviving patients required shunting (Ellenbogen et al. 1989). Two other series reported much higher rates of shunt dependency, ranging from 57 to 78% of cases reported (Humphreys et al. 1987; Lena et al. 1990). Raimondi and Gutierrez have recommended that third and fourth ventricle tumors require immediate shunt placement followed by a delay of 7–14 days prior to surgery (Raimondi and Gutierrez 1975). This method, while acceptable, can be substituted by performance of both procedures at the same time, if permitted by the condition of the patient.

Conventional catheter angiography is not required for diagnosis. Rather, its primary role is as a preop-erative adjunct to define the blood supply and can be combined with embolization to reduce tumor vascu-larity. Angiography clearly indicates that the vascular supply of papillomas is from normal choroidal vessels, which often enlarge as the tumor grows. Tumors of the lateral ventricle or third ventricle are generally supplied by branches of the anterior or posterior choroidal arteries. Mass effect tends to displace the internal occipital artery and the basal vein of Rosenthal in an inferior direction. A fourth-ventricle tumor receives its blood supply from medullary or vermian branches of the posterior inferior cerebellar artery.

9.6.2 Operative Treatment

The goal of surgery is gross total resection (GTR), as measured by postoperative MR imaging. As with most intracranial tumors, the exact approach is determined by avoiding eloquent tissue (primary motor or sensory cortex, speech centers, and visual cortex). The two features of choroid plexus tumors that can make resection exceedingly difficult are: (1) profuse vascularity and (2) large size. The tumor's arterial vessels arborize rapidly, and so control of hemorrhage within the tumor requires slow and tedious dissection. The most effective strategy focuses on initial exposure of the feeding artery and its ligation (Fig. 9.3). In general, en bloc excision is recommended (Raimondi and Gutierrez 1975). For lateral ventricle papillomas, a cerebral incision posterior to the angular gyrus allows access to the entire trigone and permits the pedicle of the tumor to be identified and coagulated. For more anteriorly located tumors, an incision can be made in the frontal convolutions and the lateral ventricle approached from an anterolateral direction. Lateral ventricle tumors can also be approached through a cerebrotomy through the superior or middle temporal gyrus.

Figure 9.3
figure 3

The same case as shown in Fig. 9.1. (a) The axial T1-weighted image clearly shows an enlarged choroidal artery leading into the tumor. (b) The postoperative coronal MRI image shows the route through the temporal lobe used to access initially the feeding artery and then the tumor itself. Once the blood supply was interrupted, the tumor removal proceeded uneventfully

Full size image

Third-ventricle tumors are rare and are approached via a midline transcallosal route. The anterior aspect of the ventricle is entered through a generous opening in the corpus callosum extending from the rostrum to the supraoptic recess. In this way, the tumor can be separated from the choroid of the tela choroi-dea where it is usually attached and the accompanying bridging vessels can be identified and divided.

Fourth-ventricle tumors almost always produce triventricular obstructive hydrocephalus, and may require preoperative shunting and stabilization as noted earlier. Tumors in this location arise from the caudal part of the roof of the fourth ventricle and may extend into the lateral recesses, or through the foramen of Magendie. The approach is via a standard midline posterior fossa craniectomy or craniotomy exposing the vermis and tonsils. The blood supply from branches of the PICA are visualized from a medial vantage.

9.6.3 Treatment of Choroid Plexus Carcinomas

Overall, reported results confirm that GTR has a favorable impact upon survival for carcinomas (see Outcome section). For this reason, aggressive surgical treatment with GTR should be the primary objective. Nevertheless, GTR with carcinoma is achieved in less than 50% of cases. Combined with adjunctive therapy, either radiation or chemotherapy, survival following GTR ranges from 67 to 91% (Fitzpatrick et al. 2002). Technical considerations with CPC include the expected increased tumor vascularity, as well as additional difficulties relating to the lack of a well-developed plane between the brain and tumor, and excessive friability of the tumor tissue. The rate of recurrence associated with GTR alone suggests that adjunctive therapy is useful, although definitive guidelines are not available (Fitzpatrick et al. 2002).

Most chemotherapy regimens rely upon cyclo-phosphamide, etoposide, vincristine, and a platinum agent (St Clair et al. 1991; Packer et al. 1992; Berger et al. 1998). Wolff et al. noted that only 8 of 22 carcinomas responded to chemotherapy, a disappointing observation (Wolff et al. 2002). Use of combination chemotherapy (ifosfamide, carboplatinum, and etoposide) after an initial surgical procedure was found to reduce tumor volume and allow a more complete resection during a second-stage operation (St Clair et al. 1991; Razzaq and Cohen 1997). Importantly, the vascularity of the tumor appeared to be greatly reduced, as measured blood loss during the second procedure was on an average, 15% of blood volume, compared to an average of 64% of blood volume during the first procedure. Recent meta-analy-ses have noted that administration of chemotherapy resulted in a survival advantage for patients with completely or incompletely resected carcinomas, and that second-look surgery is of benefit for those patients with incompletely resected CPCs (Wrede et al. 2005, 2007). Chemotherapy was also beneficial in the subgroup of patients who did not receive radiation. These observations, although retrospective in nature, suggest that aggressive therapy including chemotherapy and further attempts to remove any remaining tumor should be pursued when possible.

Postoperative radiation is usually recommended if the child is over 3 years of age, although this therapy has not been subjected to a clinical trial. Radiation is also used in the presence of leptomeningeal dissemination, subtotal resection (STR), and drop metastases. In one series, 10 patients with CPC were treated with either chemotherapy and/or cranio-spinal radiation (Chow et al. 1999). Some of these patients demonstrated no evidence of disease following chemotherapy alone, but others required radiation to achieve disease control. The authors do suggest that radiation can be used as salvage therapy, but whether radiation for all patients with carcinoma would reduce the relapse rate remains unclear. Certainly, this should be judiciously used in children under 3 years of age. Fitzpatrick et al. noted that following STR, radiation therapy, either alone or in combination with chemotherapy, offered a survival advantage (Fitzpatrick et al. 2002). The question of which adjunctive therapy to use following GTR remains unclear, although the presence of relapse despite chemotherapy and radiation suggests that surgery alone is not sufficient for CPC. Wolff et al. support this view and state that GTR alone is insufficient for carcinoma, and should be supplemented with radiation (Wolff et al. 1999). The role of confor-mal radiation and radiosurgery is unknown, nor is the role of intrathecal chemotherapy. The experience reported by Packer suggests that disease relapse confers a poor prognosis (Packer et al. 1992).

9.7 Outcome

The vast majority of patients with CPP can expect an excellent long-term survival. The survival for CPC, however, is much worse. In a recent meta-analysis, the 1-, 5-, and 10-year survival for papilloma was 90, 81, and 77%, compared to only 71, 41, and 35% for carcinoma (Wolff et al. 2002). In another large series of grade I CPPs, 12 of 124 patients recurred during a mean follow-up period of 59 months (Jeibmann et al. 2007). Of the 124 papillomas, 21 were described as having atypical histology. Six of these 21 tumors recurred, compared to 6 of the 103 tumors with normal histology. In the same series, 2 of the 103 papillo-mas progressed to carcinomas. The extent of surgery is the most important treatment variable impacting long-term survival for both papilloma and carcinoma patients (Ellenbogen et al. 1989; Packer et al. 1992; Wolff et al. 2002). The overall crude survival rate in Ellenbogen's series was 88% for patients with papil-lomas and 50% for those with carcinomas (Ellenbogen et al. 1989).

Packer et al. reported that GTR for carcinoma without adjunctive therapy offers the highest likelihood of success (Packer et al. 1992). Four of five patients who underwent GTR remained disease-free at a median of 45 months after diagnosis. Five of six patients who had a STR suffered a relapse.. Two other reports, however, noted that 5-year survival following GTR of carcinomas ranged from 26 to 40% (Berger et al. 1998; Pencalet et al. 1998). Berger et al. also noted that surgery was the most important prognostic factor for CPC. The meta-analysis by Wrede et al. confirmed the utility of chemotherapy and/or radiation for CPC (Wrede et al. 2007). A brief report noted that the 5-year survival for patients with carcinoma who were treated with GTR followed by radiation was 68%, compared to 16% for those not irradiated (Wolff et al. 1999). The two groups were not exactly comparable, but the clear suggestion is that surgery alone is insufficient to prevent recurrence of carcinomas.

Although papillomas are histologically benign and potentially curable, morbidity and mortality are sig-nificant concerns. With respect to operative mortality, modern series provide figures of 8–9.5% (Humphreys et al. 1987; Lena et al. 1990). In the series from the Hospital for Sick Children the cumulative mortality was 36%, the majority of which (6 of 8) occurred in patients below 12 months of age. Morbidity remains an important problem. In one series 33% of patients with papillomas had persisting motor sequelae and psychomotor retardation (Lena et al. 1990). In another series, 26% of patients were classified as having a fair or poor recovery (Ellenbogen et al. 1989).

As noted earlier, the treatment of hydrocephalus goes hand-in-hand with the treatment of choroid neoplasms, and associated complications can occur. One significant complication is the presence of large sub-dural collections that may develop following tumor resection, caused by a persistent venticulosubdural fistula. Boyd and Steinbok appear to have dealt with this problem by applying pial sutures at the conclusion of the procedure (Boyd and Steinbok 1987). The role of preoperative shunting in the causation of this entity is unclear.

9.8 Conclusions

Choroid plexus tumors represent a well-defined subset of brain tumors that occur mainly in young children. Surgical resection for papilloma is usually curative, while adjunctive therapy for carcinoma should include chemotherapy and/or radiation. The long-term survival for carcinoma remains poor. The overall functional outcome can be excellent, but the potential for neurologic morbidity should be recognized early even for benign tumors.