Clinical Features

Primary choroid plexus neoplasms are found in patients of any age, from birth to the eighth decade of life. More than 70% occur in children, and at least 50% are found in children younger than 2 years.3,9,27–30 Reports of in utero tumors by prenatal ultrasonography seem to indicate a congenital origin for some of these tumors.31,32 A preponderance of male patients has been described in several series3,33–37; however, other series show a fairly even distribution between males and females.6,38 There do not seem to be differences in age at diagnosis, sex, symptoms, or location of tumor in patients with papilloma as opposed to those with carcinoma.3

The patient′s initial clinical features can be grouped into four overlapping sets of signs: increased intracranial pressure (ICP), seizures, hemorrhage, and focal neurologic abnormalities. By far, most children are brought to the hospital for an investigation of hydrocephalus or macrocephaly. These patients present with signs of increased ICP, which include tense fontanelle, splayed sutures, vomiting, lethargy, and irritability.3,9,33 The symptoms of choroid plexus tumors are caused predominantly by the hydrocephalus that afflicts approximately 90% of pediatric patients, and the duration of symptoms ranges from several weeks to 6 months. Although most infants present with intracranial hypertension, the presentation in young children may be more nonspecific. Less common but well-documented presentations include head tilt, titubation, weakness, failure to thrive, developmental delay, and “shunt-resistant” hydrocephalus.39–43 The shunt-resistant hydrocephalus is secondary to a choroid plexus tumor, which can produce many times the normal rate of cerebrospinal fluid (CSF).44 This overproduction of CSF can be sufficient to cause abdominal ascites if a ventriculoperitoneal shunt is placed prior to surgical removal of the tumor.43 Although rare, overproduction of CSF can be up to 5 L per day as reported in one case. This patient had successful control of CSF production by a one-stage bilateral choroid plexotomy, which revealed a choroid plexus papilloma.45 Even such rare side effects as psychosis and bobblehead-doll syndrome have been reported in the literature ( Table 24.1 ).46

Children with posterior fossa lesions may exhibit signs of brainstem compression, cranial nerve paresis, and cerebellar dysfunction. The scarcity of tumors in the cerebellopontine angle in children makes it difficult to generalize on their presentation, although hydrocephalus is a prominent finding ( Table 24.2 ).47–49

Children with third ventricular tumors present with hydrocephalus but may also exhibit endocrine disturbances, precocious puberty, diabetes insipidus, or diencephalic disorders (Table 24.3 ).23,50,51

An interesting feature of choroid plexus tumors is their anatomic distribution. They are most commonly found in the lateral ventricles in children and less commonly in the fourth ventricle, in contrast to adults, in whom the majority of these tumors are found in the fourth ventricle and its lateral recesses.2,28,29,37,52–55 No developmental feature adequately explains this unique anatomic distribution. Approximately 75% of tumors in children are found in the lateral ventricles. Most tumors in the lateral ventricle are found in the atrium/trigone, but they may also be seen in the temporal horn or near the foramen of Monro. Van Wagenen′s22 original report that the majority of lateral ventricular tumors are found in the left ventricle has not been supported by other series, which find a similar number in right and left lateral ventricles.3 Location in the third ventricle is less common but has been described in both children and adults.37,38,43,56 Third-ventricular choroid plexus neoplasms in pediatric series average approximately 10%3,51,56 and range from 0 to 29%.33 Primary extraventricular locations are the rarest sites, but tumors have been described in children in the cerebellopontine angle, suprasellar cistern, foramen magnum, and spinal subarachnoid space.2 Extension of the tumor from one ventricle to another, or through a ventricular foramen, or into a subarachnoid cistern is occasionally noted.29,43 In Rovit et al′s57 1970 review of 245 cases of choroid plexus tumor reported in the literature, nine cases (3.7%) of tumor in multiple locations were discovered, six of which were in the lateral ventricle. However, since the advent of advanced imaging techniques, these tumors can be visualized more accurately extending into other CSF spaces and the brain parenchyma. Choroid plexus tumors in children, like other intraventricular tumors in adults, are often quite large by the time of diagnosis; in one series, they averaged 4 cm in diameter at the time of diagnosis ( Table 24.4 ).3

Hydrocephalus, which is a cardinal manifestation of choroid plexus neoplasms, can be the result of obstruction of CSF pathways or overproduction of CSF. Overproduction of CSF by these tumors has attracted much interest and been the subject of numerous reports.20,29,44,58–64 Initially, the resolution of hydrocephalus after complete resection of choroid plexus tumors was the basis for postulating that oversecretion of CSF was responsible for the hydrocephalus.6 However, an increased rate of CSF production has been unequivocally demonstrated with choroid plexus neoplasms.28,44,61,63–65 Rekate et al66 demonstrated that although overproduction of CSF can cause hydrocephalus, a complex combination of CSF overproduction and limited outflow may be the cause in many patients.

Tumor mass, blood products, tumor products, cellular debris, and metastases can mechanically obstruct the normal egress of CSF, thus causing hydrocephalus. Necrosis, tumor hemorrhage, operative hemorrhage, and ependymitis have all been implicated as causes of hydrocephalus in patients with such choroid tumors.36,40,49,67–70 Elevation of CSF protein is found in the majority of patients, and xanthochromia in the CSF is common; frank hemorrhage is less common.6 The obstruction of CSF absorptive pathways may also explain why, despite a gross total excision of tumor, the hydrocephalus may persist. Approximately one third to one half of patients with choroid plexus neoplasms require permanent CSF diversion postoperatively even when the tumor is completely removed.3,33,49,68,69

Another interesting entity reported is diffuse villous hypertrophy of the choroid plexus, a term originally coined by Davis16 in 1924 to describe a bilateral choroid plexus lesion. As there is little histopathological difference between papilloma and normal choroid plexus tissue, it is difficult to determine if this entity represents a true neoplasm or simple hypertrophy of the choroid plexuses. Bilateral papilloma is a rare condition, as demonstrated by its absence from series of choroid plexus papilloma.6,27,29,53–55 The observation that overproduction of CSF can result from villous hypertrophy of the choroid plexus or papilloma has since been reported by several investigators who commented on the difficulty in diagnosis.16,28,40,64,71,72 Placement of a ventriculostomy and measuring CSF formation may help in the documentation of overproduction of CSF. Laurence28 described six tumors among his own patients but was cautious about their removal. However, several surgeons have shown that resection of these rare bilateral choroid plexus lesions results in resolution of the CSF overproduction and alleviation of the concomitant hydrocephalus.64,71,72

Imaging

Plain films may show marked calcification and nonspecific signs of increased pressure, such as splayed sutures.6,27,73 Historically, pneumoencephalography, ventriculography, and angiography were utilized to show these tumors to be intraventricular masses with associated hydrocephalus.6 Unfortunately, several deaths were associated with the application of pneumoencephalography and ventriculography in the diagnosis of choroid plexus tumors as well.28,49 These deaths were probably a result of the large shifts created by placing a brain needle with air or fluid into large ventricles with a large mass.

The introduction of computed tomography (CT) improved the safety and accuracy of diagnosis, and ultimately the outcome, in children harboring these tumors.29 Both papilloma and carcinoma appear isodense to hyperdense on CT scan, with frequent calcification and usually marked enhancement. Tumors are spherical, multilobular, and sometimes cystic. They may extend to another ventricle or CSF cistern, a finding not entirely specific for choroid plexus tumors ( Figs. 24.1 and 24.2; Table 24.5 ).

Magnetic resonance imaging (MRI) usually shows a mass that is isointense or slightly hypointense to gray matter on T1-weighted images and hyperintense on T2-weighted images. This tumor is often a lobulated, homogeneous, intraventricular mass on both short repetition time/echo time (TR/TE) and long TR/TE sequences.74 Enhancement with paramagnetic substances, such as gadolinium-diethylenetriamine pentaacetic acid (DTPA), is markedly bright and usually homogeneous, although various patterns can be seen, including nodular, peripheral, and cyst wall enhancement. A common finding is the presence of serpentine signal voids, indicating that an enlarged blood vessel is supplying the tumor.75,76 MRI is the diagnostic imaging study of choice because of its detailed anatomic delineation and triplanar imaging ability. Magnetic resonance angiography may provide additional information regarding the vascular supply of the tumor. The relative ease and availability of magnetic resonance angiography has slowly supplanted the need for transfemoral cerebral angiography in the management of these tumors in children, unless presurgical embolization is required ( Figs. 24.2 and 24.3; Table 24.6 ).

Both papilloma and carcinoma can compress the surrounding brain, although brain invasion is a characteristic of carcinoma.77 Some authors have commented on the peritumor vasogenic edema in the surrounding white matter in carcinoma.75,77 Carcinoma does have a greater tendency to invade the brain, but it can also be found in an entirely intraventricular location. The radiological differential diagnosis of choroid plexus tumors in children includes ependymoma, primitive neuroectodermal tumor, astrocytoma, germinoma, teratoma, meningioma, and metastases to the choroid plexus. Despite the significant difference in histological appearance, these tumors can appear homogeneous on MRI and quite similar in magnetic resonance appearance to choroid plexus tumors. Other unusual tumorlike masses that can appear similar include inflammatory pseudotumor, choroid plexus cysts, and xanthogranulomas ( Table 24.7 ).

Pathology and Differential Diagnosis

On gross analysis, a choroid plexus papilloma has a cauliflower-like surface and appears as a generally well-circumscribed mass arising from within a ventricle. The papilloma can extend into the brain parenchyma, compressing it along a broad margin. Heavily calcified tumors may be difficult to section without first being decalcified.2 Microscopically, choroid plexus tumors typically appear as a single layer of cuboidal epithelial cells surrounding a delicate fibrovascular stalk, arranged in a papillary configuration with fingerlike projections of tissue. A well-formed continuous basement membrane is a prominent feature noted in all cases. Choroid plexus tumors span the histological spectrum from extremely well-differentiated tumors to anaplastic tumors with minimal epithelial differentiation. Both well-differentiated and poorly differentiated components may be seen in a single tumor.2 However, the majority of choroid plexus neoplasms lie in the well-differentiated part of the spectrum. It is occasionally difficult to differentiate normal choroid plexus from a papilloma, but the latter contains cells that are more crowded, columnar in shape, pleomorphic, and demonstrating more variation in nuclear size. The nuclear/cytoplasmic ratio is often increased ( Fig. 24.4 ).2

Ultrastructural observations include apical microvilli, scattered cilia, and interdigitating lateral cell borders sitting atop a basement membrane that is seated on a delicate fibrovascular stalk. Stromal calcification or xanthomatous changes may be evident.2 Atypical microscopic features may be observed in papilloma; these include increased cellularity (two or three cell layers as opposed to one), mitoses, nuclear pleomorphism, and poorly formed papillary structures. These intermediate tumors may possess one or two of these features and are called atypical papilloma, but do not necessarily have a more aggressive natural history and are not classified as carcinoma.2 However, anaplastic transformation of well-differentiated tumors has been reported to occur over time. Although uncommon, a lesion that is initially well differentiated or atypical can become anaplastic as documented during a subsequent resection.78,79

The histological features of papilloma versus carcinoma have been the subject of both close scrutiny and controversy. The focus in diagnosis often centers on this important question: What characteristics differentiate papilloma from carcinoma? Review of the literature reveals that from a historical perspective the criteria for differentiation of papilloma from carcinoma have not been uniform. The definition of choroid plexus carcinoma has varied slightly in several studies based on histological criteria.3,11,36,80,81 Dohrmann and Collias,11 utilizing Russell and Rubinstein′s36 and Lewis′s82 criteria of carcinoma, reviewed 22 cases of primary choroid plexus carcinoma in the literature from 1844 to 1975 and found that only 11 satisfied the criteria of carcinoma. Similarly, Lewis′s review of the literature resulted in the dismissal of many other claims of carcinoma.

The consistent histological features of choroid plexus tumors that unequivocally differentiate carcinoma from papilloma include cellular anaplasia, loss of differentiated papillary choroid architecture, nuclear pleomorphism, mitosis, necrosis, and giant cell formation ( Fig. 24.5 ). At the extreme end of the spectrum are the sheets of anaplastic cells without appreciable papillae. These tumors are very cellular with complex cribriform structures, with a high mitotic index. One issue of contention centers around whether brain invasion either is required or by itself secures the diagnosis of carcinoma. Some authors insist that invasion be demonstrated histologically, whereas others do not.81 It is noteworthy that not all surgical specimens have brain tissue insinuated within their projections, especially in cases where the surgeon used ultrasonic aspiration or microsurgical suction to remove the specimen at the brain–tumor interface and brain is not included in the surgical specimen. Thus, the absence of fingerlike projections invaginating into the brain parenchyma does not necessarily eliminate the diagnosis of carcinoma. Some tumors may invade the stroma, but on careful histopathological analysis are shown not to have invaded the brain. In addition, some choroid tumors with relatively benign histological features are invasive, whereas some anaplastic tumors seem to have circumscribed borders.3,83 Often, the diagnosis of invasion is inferred by MRI, not by histological evidence. Thus, although invasion of brain makes the diagnosis of carcinoma highly likely, such diagnosis is not invariably secured unless there are associated malignant cellular features ( Fig. 24.5; Table 24.8 ).

Some authors maintain that even in choroid plexus papilloma macro- and microscopic implants can be found in the leptomeninges, the subarachnoid space of the spinal cord, and the ventricular system.84,85 A rare case of pulmonary metastases from a choroid plexus papilloma was reported by Vraa-Jensen86 in 1950 in an 11-year-old girl. The child developed pulmonary metastasis and lesions in the skull, which demonstrated malignant transformation (carcinoma) at autopsy. Clinically symptomatic metastases from papilloma are extremely rare.

The electron microscopic findings are useful for distinguishing choroid plexus carcinoma from ependymoma, which can occasionally share a close histological appearance. The presence of a basal lamina as seen in choroid plexus tumors often rules out ependymal origin of a tumor.87

Some studies have demonstrated the diagnostic utility of immunohistochemical methods. Immunohistochemical staining of choroid plexus tumors reveals both epithelial and glial characteristics. Choroid neoplasms are positive for cytokeratin (epithelial), S100 (diffuse staining), and vimentin.88,89 Choroid plexus carcinoma retains cytokeratin positivity but decreased S100 staining.79 Positivity for carcinoembryonic antigen is more common in carcinoma, whereas S100 positivity is more common in papilloma.90 Ependymoma is characterized by nonepithelial glial elements as well as epithelial components. Ependymomas lack the prominent basement membrane seen in choroid plexus tumors and seen on periodic acid–Schiff (PAS) preparations, electron micrographs, or immunohistochemical specimens stained for laminin.91 Choroid plexus neoplasms are positive for cytokeratin, whereas ependymomas are not; ependymomas are widely glial fibrillary acidic protein (GFAP) positive, whereas choroid plexus tumors are diffusely but not uniformly GFAP positive. In an adult with an intraventricular mass, S100 positivity favors a choroid plexus primary tumor versus a metastatic tumor ( Table 24.9 ).2

In an immunohistochemical study of choroid plexus neoplasms, it was noted that a child who survived with a choroid plexus carcinoma had a tumor that stained positive for S100 and negative for epithelial membrane antigen, as often seen in papilloma. This immunohistochemical profile was not present in another child in the series, who died with carcinoma.92 This outcome and staining profile suggested a biological or genetic variability in carcinoma that may in the future help us identify which children are likely to do better than others. In addition, DNA sequences similar to those found in simian virus 40 (SV40) have been found in choroid plexus tumors. This provocative finding is supported by the observation that some transgenic mice infected with SV40 have been known to develop choroid plexus neoplasms, which is indicative of the oncogenic properties of polyoma viruses.93 However, in summary, no clear-cut immunochemical or molecular criteria have emerged with any consistency to separate papilloma from carcinoma or provide for prognostic significance.