Choroid plexus tumours: Classification, MR imaging findings and pathological correlation


  • C Yan MD; Y Xu MD, PhD; J Feng MD, PhD; C Sun MD, PhD; G Zhang MS; J Shi MS; P Hao MD; Y Wu MD, PhD; B Lin MD.
  • Conflict of interest: None.


Professor Yikai Xu, Department of Medical Imaging Center, Nanfang Hospital, Southern Medical University, no. 1838 Guangzhou Avenue North, Guangzhou 510515, Guangdong, China.




Choroid plexus tumours (CPTs) are extremely rare intraventricular neoplasms and are prone to bleeding during surgery. The purpose of this study was to summarise the MR imaging characteristics of 13 CPT cases.


Magnetic resonance images of 13 patients (six men and seven women; mean age 21.1 years) with pathologically proved CPTs were retrospectively reviewed. MR findings of the tumours were evaluated, with emphasis on their location, size, shape, internal architecture, margin and pattern and degree of enhancement. Differences in signal intensity characteristics were also investigated on MR images and analysed according to histological subtypes.


Lesions were in the lateral ventricles (n = 7), fourth ventricle (n = 5) and cisterna magna (n = 1), with a mean size of 5.0 cm (range 2.0–7.9 cm). The tumour parenchyma was a mixture of nodular or patchy areas of inhomogeneous isointense to slightly hyperintense signal on T2-weighted images. On postcontrast MR images, all lesions, except for one, had moderate to marked contrast enhancement. Multiple tortuous areas of ‘flow void’ signal extended through all the tumours except for two. A thin capsule could be seen in six cases.


Observation of large intraventricular tumours with inhomogeneity on T2-weighted images and flow void is suggestive of CPTs. Checking for signs of a thin capsule, extensive peritumoural oedema and necrosis may be useful when classifying CPTs.


Choroid plexus tumours (CPTs) are rare intraventricular neoplasms derived from the choroid plexus epithelium and account for only 0.4% to 0.6% of all intracranial neoplasms.[1] The average annual incidence of CPTs is approximately 0.3 per 1 000 000 people.[2] According to the 2007 WHO Classification of Tumours of the Central Nervous System, [3]primary CPTs are classified as choroid plexus papilloma (CPP, WHO grade I), atypical CPP (WHO grade II) and choroid plexus carcinoma (CPC, WHO grade III). Atypical CPP, which is defined as a new subtype of CPTs, has few histological features of malignancy such as increased mitotic activity, necrosis and nuclear pleomorphism.[4, 5] However, because of the rarity and imaging diversity of the CPTs, accurate diagnosis and classification is difficult.

A non-invasive method to improve preoperative diagnosis is important to help determine the most appropriate surgical plan because CPP can be removed completely; however, many patients with atypical CPP or CPC tumours require a more extensive surgery to avoid recurrence and metastases.[6] Surgical protocols are substantial to avoid intraoperative, uncontrolled bleeding owing to the extreme vascularity of CPTs. Although a limited number of case reports or small series are focused on the pathological course, clinical course and radiological characteristics,[5, 7-9] no literature exists that systematically summarises the MR imaging features according to the 2007 WHO classification of CPTs. Therefore, this paper mainly documents the MR findings in 13 pathologically confirmed CPTs, aiming to improve accurate preoperative diagnosis and classification.



Between April 2004 and May 2012, the findings in 13 patients with histologically proven CPTs (as determined by examining department of pathology records) were retrospectively reviewed. The patients were recruited after a review of medical records from four tertiary referral academic centres. There were six males and seven females, aged from 5 months to 58 years (mean 21.1 years; median 18 years). The indications for scanning were headache (n = 8), vomiting (n = 6), dizziness (n = 2), weakened physical strength (n = 2), bulging fontanelle (n = 1) or behaviour disorder (n = 1). The duration of symptoms before diagnosis was from 10 days to 17 years (median 1 month). Ethics approval for this retrospective record review was obtained from our Medical Ethics Committee.


Preoperative MR images of all patients were available. Six of the patients were preoperatively imaged on a 1.5T (MAGNETOM Vision Plus, Siemens, Erlangen, Germany), four on a 1.5T (Achieva, Philips, Best, The Netherlands) and three on a 3.0T (Signa EXCITE, General Electric, Milwaukee, Wisconsin, USA). Precontrast T1-weighted imaging (T1WI) fast spin echo (FSE) images (repetition time (TR) 450–600 ms; echo time (TE) 14–16 ms) and T2-weighted imaging (T2WI) FSE images (TR 5000–5100 ms; TE 99–137.1 ms) of all patients were obtained. Fluid-attenuated inversion recovery (FLAIR) images (TR 8000 ms; TE 120 ms; inversion time 110 ms) of 10 cases were available. The axial, coronal and sagittal contrast-enhanced T1WI were obtained after intravenous injection of 0.1-mmol/kg of Gd-DTPA. The thickness of the layer was 5–10 mm.

All images were independently reviewed by two experienced radiologists to evaluate the location, size, shape, internal architecture, margin and pattern and degree of enhancement of the lesion. The size of the lesion was measured at the greatest diameter of the lesion. On postcontrast MR images, the pattern of enhancement at the solid portions of the lesion was categorised as homogeneous or heterogeneous. The degree of enhancement was also subjectively assessed as being poor, mild, moderate or marked.


The MR imaging findings and clinical features are shown in Tables 1 and 2. The lesions were in the lateral ventricles (n = 7), in the fourth ventricle (n = 5) and within the cisterna magna (n = 1). The diameter of the lesions ranged from 2.0 to 7.9 cm (median 5.0 cm). All lesions were either rounded (n = 3) or lobulated with a surrounding cauliflower rim (n = 10). Compared with the grey matter, 12 lesions (12/13) were slightly hypointense on T1WI, and the remaining lesion had mixed hypointensity and hyperintensity. On T2WI, the tumour parenchyma was markedly inhomogeneous (mixed with nodular or patchy areas of both isointense and slightly hyperintense signal intensity) (Figs 1b,3b). Ten cases had heterogeneous hypointense and isointense regions on the FLAIR images (Fig. 2c). Scattered intratumoural necrotic and cystic areas were observed in four cases, with an additional surrounding haemosiderin ring in one case (Fig. 2). Foci or nodular calcifications were seen in three cases (patients 2, 5 and 12), later confirmed by examining pathological sections. Multiple tortuous areas of ‘flow void’ signal extended through all the tumours except for two (Figs 1-3). All lesions but one had moderate to marked contrast enhancement on postcontrast images. Relatively homogeneous enhancement was observed in 11 lesions (Figs 2d,3b). Marked enhancement occurred at the slightly hyperintense areas, and mild to moderate enhancement was seen at the isointense areas on T2WI. A total of 7 of the 13 patients had mass effect with severe dilation of adjacent ventricles. Additionally, margins of lesions were well defined in 11 cases, six of which had an isointense capsule (Fig. 1a,b), but margins were partly indistinct in two cases. Moreover, marked peritumoural oedema was observed in four cases (Fig 3a). All lesions were resected by open surgery, and pathological examination was performed for all cases. During surgical procedures, the lesions were seen as relatively well circumscribed and highly vascular masses with a dark or reddish-grey surface. Most lesions (n = 8) were classified as CPP (Fig. 1); the remaining were atypical CPP (n = 3; Fig. 2) or CPC (n = 2; Fig. 3).

Figure 1.

Case 1: Choroid plexus papilloma (CPP) in a 58-year-old male (WHO grade I). Magnetic resonance images show that the tumour is located in the right lateral ventricle with mild mass effect. The tumour parenchyma is hypointense to the grey matter on T1-weighted imaging (T1WI) (a) and isointense on T2-weighted imaging (T2WI) (b), with inhomogeneous enhancement on Gd-enhanced T1WI (c). Speckled signal voids (arrows) and a thin isointense capsule can be seen on T1WI and T2WI. The photomicrograph of CPP (d) shows a papillary neoplasm composed of fibrovascular cores covered by a single layer of cubodial to columnar epithelium (haematoxylin–eosin × 400).

Figure 2.

Case 9: Atypical choroid plexus papilloma in a 15-year-old female (WHO grade II). Magnetic resonance images show that the tumour is located in the fourth ventricle with mild mass effect and significant peritumoural vasoedema. The tumour parenchyma is homogeneous hypointense to the grey matter on T1-weighted imaging (T1WI) (a) and inhomogeneous isointense on T2-weighted imaging (T2WI) (b) and fluid attenuated inversion recovery (c), with marked enhancement on Gd-enhanced T1WI (d). Tortuous signal voids (arrows) and ring-shaped hyperintense of haemorrage (arrowheads) can be seen on T1WI and T2WI.

Figure 3.

Case 12: Choroid plexus carcinoma in an 18-year-old female (WHO grade III). Magnetic resonance images show that the tumour is located in the obviously dilated right lateral ventricle with mild peritumoural vasoedema. The tumour parenchyma is isointense to the grey matter with patchy areas of necrosis (arrowhead) on T2-weighted imaging (a). Speckled signal voids (shorter arrow) can be seen on Gd-enhanced T1-weighted imaging (b).The low signal (longer arrow) on all sequences is calcification demonstrated by the pathological section (a–c).

Table 1. Summary of MR imaging features of 13 patients with choroid plexus tumours
PatientLocationSize (cm)ShapeT1WIT2WIFlow voidCystic necrosisCapsuleEnhancement pattern/degreePeritumoural oedema
  1. L, left; R, right; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging.
 1 (Fig. 1)R lateral ventricle5.8LobulatedSlightly hypointenseInhomogeneous isointenseYesNoYesHomogeneous moderateMinimal
 2Fourth ventricle5.3LobulatedSlightly hypointenseInhomogeneous isointenseYesNoNoHomogeneous markedMinimal
 3L lateral ventricle7.9LobulatedSlightly hypointenseInhomogeneous hyperintenseYesNoYesHeterogeneous markedMinimal
 4Cisterna magna3.7RoundedSlightly hypointenseInhomogeneous hyperintenseYesNoYesHomogeneous mildModerate
 5Fourth ventricle2.0LobulatedSlightly hypointenseInhomogeneous isointenseNoNoYesHomogeneous moderateMinimal
 6R lateral ventricle5.5LobulatedSlightly hypointenseInhomogeneous isointenseNoYesYesHomogeneous markedMinimal
 7Fourth ventricle3.8LobulatedSlightly hypointenseInhomogeneous hyperintenseYesNoNoHomogeneous markedMinimal
 8Fourth ventricle3.5LobulatedSlightly hypointenseInhomogeneous hyperintenseYesNoNoHomogeneous moderateMinimal
 9 (Fig. 2)R lateral ventricle5.5LobulatedSlightly hypointenseInhomogeneous hyperintenseYesYesNoHomogeneous markedMarked
10R lateral ventricle6.6LobulatedSlightly hypointenseInhomogeneous hyperintenseYesNoYesHomogeneous markedMarked
11Fourth ventricle6.0RoundedHeterogeneous hypointenseInhomogeneous isointenseYesNoNoHeterogeneous moderateMinimal
12 (Fig. 3)R lateral ventricle5.0RoundedSlightly hypointenseInhomogeneous hypointenseYesYesNoHomogeneous moderateMarked
13R lateral ventricle5.8LobulatedSlightly hypointenseInhomogeneous isointenseYesYesNoHomogeneous markedMarked
Table 2. Demographics, surgery, follow-up and pathology of 13 patients with choroid plexus tumours
PatientSexAgeBlood loss (ml)Duration of symptomsClinical presentationExtent of tumour removalFollow-upLocal recurrenceMetastasesPathology
  1. a, alive; CPC, choroid plexus carcinoma; CPP, choroid plexus papilloma; CR, complete resection; d, dead; F, female; m, months; M, male; SR, subtotal resection; w, weeks; y, age in years.
 1M58 y4006 monthsHeadache, vomitingCR11 m (a)NoNoCPP
 2F26 y25015 daysHeadache, vomitingCR8 m (a)NoNoCPP
 3M5 m8005 monthsBulging fontanelleCR0 (d)NoNoCPP
 4F35 y3001 monthHeadache, vomitingCR75 m (a)YesNoCPP
 5F28 y20020 daysDizzinessCR8 w (a)NoNoCPP
 6M20 m3001 monthWeakened physical strengthCR6 w (a)NoNoCPP
 7M29 y4006 monthsHeadache, dizzinessCR31 m (a)NoNoCPP
 8M21 y30010 daysHeadacheCR3 w (a)NoNoCPP
 9F15 y250020 daysHeadache, vomitingCR0 (d)NoNoAtypical CPP
10F42 y50017 yearsHeadache, vomitingCR65 m (a)NoNoAtypical CPP
11M8 m2002 weeksWeakened physical strengthCR7 w (a)NoNoAtypical CPP
12F18 y4004 monthsHeadache, vomitingSR44 m (a)NoNoCPC
13F11 m8001 monthBehavioural disorderCR13 m (a)NoNoCPC


Epidemiological, pathological and clinical characteristics of CPTs

CPTs are uncommon neoplasms that arise from specialised secretory cells lining the choroid plexus and may originate wherever a choroid plexus is found within the ventricular system.[10]CPTs are most common in the lateral (54%) and fourth ventricles (26%), with some specimens occurring in the third ventricle (11%) or other rarer locations such as the cerebellopontine angle,[11] consistent with the distribution in our study. CPTs may manifest at any age but predominantly occur in children younger than 2 years, accounting for 2% to 3% of intracranial neoplasms in children and approximately 0.6% in adults.[1, 12] CPP outnumbers CPC by a ratio of 5:1, and approximately 80% of CPC occurs in children.[5] However, the number of adults with these tumours might be underestimated.[13] Generally, there is a slight male predilection: the male-to-female ratio is 1.22:1.[6] Contrary to most reports of findings, the majority of our patients were adults (61.5%) and female (53.8%), which may be related to statistical fluctuations in small samples.

In routine microscopy, CPP resembles the normal choroid plexus, with crowding and elongation of cells and some cellular atypia in the papilloma. A typical configuration of CPTs consists of fibrovascular cores lined by a single layer of cuboidal to columnar epithelioid cells that are embryologically derived from the ependyma.[14] The fourth edition of the WHO Classification of Tumours of the Central Nervous System, published in 2007, identified atypical CPP as a neoplasm that is intermediate in histology and graded between CPP and CPC. Additionally, atypical CPPs have only one or a few atypical histological features, including increased mitotic activity, pleomorphism, distortion or blurring of the papillary structure and necrosis but are not unequivocally malignant.[15, 16] Jeibmann et al.[4] reviewed 164 cases of CPT and proposed defining atypical CPP by mitotic activity (>2 mitoses per 10 high-power fields) to distinguish it from CPP. The final diagnosis and classification of CPTs may depend on immunohistochemistry characteristics such as the expression of the S-100 protein, vimentin and epithelial membrane antigens.

Caused by the overproduction of cerebrospinal fluid and the mechanical obstruction of cerebrospinal fluid pathways, CPTs tend to result in hydrocephalus and signs of increased intracranial pressure.[17] The symptoms and signs of increased intracranial pressure include headache, vomiting, seizures or behavioural changes in adults and bulging fontanelles, delayed development and vomiting in children. Headache and vomiting were the most common symptoms in our study. Additionally, dizziness, bulging fontanelle, weakened physical strength and behaviour disorder were found. The duration of symptoms in our series was variable. However, the symptoms and signs caused by CSF obstruction are non-specific in CPTs, so it is difficult to differentiate them from other ventricular tumours by clinical features.

The most important prognostic factor is complete surgical excision.[18] Most CPTs are benign in nature and can be cured by surgery, so complete excision should be the objective of any treatment protocol.[8, 18] In atypical CPP, curative operation is possible, but the recurrence rate is much higher.[19] Notably, gross total resection is risky due to the uncontrollable intraoperative blood loss in patients with highly vascular tumours. Large quantities of blood should be prepared for transfusion during surgery. In our series, one patient died perioperatively due to a very poor general condition, and another patient died as a result of uncontrollable intraoperative bleeding. An early diagnosis of malignancy may give a better chance for cure. In CPP, complete resection is more frequently achieved (80.4%) than in atypical CPP (61.5%) or CPC (39.6%).[6] In our study, all patients, with the exception of one, underwent complete resection, with subtotal resection performed in one case because of local invasion of the tumour into vital structures. CPC carries a dismal prognosis, especially after subtotal resection.[20] CPC grows rather aggressively, with a 5-year survival rate of 26–40%, yet CPP has a more favourable outcome, with a 5-year survival rate approaching 100% and occasional recurrences. Earlier metastases and higher recurrence rates are associated with cellular atypia and increased mitotic activity.[21] Regardless of various subtypes, the histological appearance of the tumour may not correlate with the biologic behaviour. As in case 4 in our series, recurrence may occur in the benign histological subtype of CPTs. MR imaging can detect metastases, and recurrence should be monitored at an early stage, which may be useful for follow-up of patients with CPTs.

Magnetic resonance imaging findings and pathological correlation of CPTs

MR imaging findings aid clinicians in making a preoperative diagnosis when intraventricular neoplasms are suspected. In the present study, MR imaging provided important information about the characteristics of the lesions, allowing correct diagnosis and classification. Because of their origin in non-eloquent sites, most CPTs reach a considerable size (median 5.0 cm) before radiologic examination.[9] Sagittal and coronal images were valuable in assessing the amount of intraventricular tumour and route of extension. On T2WI images, the tumour parenchyma was markedly inhomogeneous (mixed with nodular or patchy areas of intermediate-to-high signal intensity), while the slightly hyperintense areas on T2WI showed marked enhancement on postcontrast images. This sign is most likely due to the crowding and elongation of cells and some cellular atypia in the papilloma. This specific feature is a helpful diagnostic sign of CPTs. To our knowledge, this characteristic has not been described previously. Solitary or multiple areas of low signal were also seen within the soft tissue mass on T2WI, which may be due to high cellularity, calcification or haemorrhage. On T1WI, most lesions in our patients were of homogeneous hypointensity, and the remaining lesion showed heterogeneous signal intensity, with surrounding ring-like hyperintense haemorrhage. Findings compatible with subacute haemorrhage within the neoplasm were verified in the pathology report, which is in part due to the fragility of dilated and congestive vessels. Speckled, stippled or tortuous areas of signal void in the central areas of tumours, interpreted as vascular proliferation and ectasis, were seen in all of our patients except for two. This feature was an additional characteristic sign of CPTs in our series, as reported in the literature.[9] Consequently, scattered areas of cystic necrosis, with uniform signal intensity equivalent to that of CSF on all sequences, were more likely to occur in the peripheral area of the tumour. This finding was noted in most cases of atypical CPP or CPC but in only one case of benign subtype, in accordance with other studies showing that cystic necrosis is uncommon in CPP.[5] According to histopathological evidence, cystic necrosis can be used as an indicator of malignancy. Intratumoural calcification has been reported in 24% of CPTs,[22] whereas foci or nodular calcifications were seen in 23.0% of the cases in our series. A thin, transparent capsule, appearing as isointensity on both T1WI and T2WI, was seen in most cases of CPP (5/8) in our series. In contrast, no case of the malignant subtype had this sign. Consequently, having an isointense capsule on T2WI may be a hallmark of benign CPTs, which has not been reported by other investigators. Significant peritumoural vasogenic oedema (Fig. 3a), which suggests penetration of the ventricular wall and parenchymal infiltration,[16] was noted in four cases of atypical CPP or CPC. Invasion of the adjacent parenchyma and surrounding cerebral oedema has been suggested to be characteristic of CPC[9, 12] but was also seen in one case of CPP (patient 4). Tumour margins with adjacent parenchyma were well delineated and defined in 11 patients and moderately well delineated in two. On T2WI, boundaries of the neoplasms were partly obscured by cerebrospinal fluid, with dilated ventricles and surrounding parenchymal oedema (patients 3 and 10). On postcontrast MR images, the tumours had more distinct margins for enhancement. Marked, homogeneous enhancement is thought to be a characteristic feature of CPTs.[23] However, we found no relationship between the variable degree and patterns of enhancement and the histological subtypes of CPTs in our series. Implantation or seeding within the subarachnoid space may occur in all subtypes of CPTs but was not observed in any of our cases. Distant metastases of CPC to other intracranial locations may occur;[8, 12] however, no metastases were observed in our study with a mean follow-up of 22.9 months.

Differentiation from other intracranial tumours

Before one suggests the diagnosis of CPTs, several other tumour types should be ruled out. The most common intraventricular tumours compared with CPTs at different sites are shown in Table 3. The signal intensity of ependymomas tends to mimic that of CPTs; however, most CPTs occur in lateral ventricles, and ependymomas are more commonly infratentorial and often have cystic components.[24] Compared with CPTs, meningiomas tend to show isointensity on T1WI and T2WI, with relatively homogeneous enhancement. The peak age for the incidence of meningiomas is older than 30 years.[25] Medulloblastomas occur predominantly in the posterior fossa. The MR features of the tumour in adults are variable, showing mild to moderate enhancement after the injection of contrast medium and having cystic and necrotic degeneration.[26] Radiologically distinguishing between medulloblastoma and CPC can be difficult when marked peritumoural oedema occurs. Central neurocytomas are more heterogeneous than CPTs on T2WI because of multiple small cysts and calcifications, which often originate near the foramen of Monro and are attached to the septum pellucidum or ventricular wall.[27, 28] Less likely lesions, including subependymal giant cell astrocytoma, lymphoma, metastases and colloid cyst, should also be considered.

Table 3. Differential diagnosis of choroid plexus tumours
LocationDifferential diagnosis
Lateral ventricleMeningioma, metastases, central neurocytoma, subependymal giant cell astrocytoma, lymphoma
Third ventricleColloid cyst, meningioma, ependymoma
Fourth ventricleMedulloblastoma, ependymoma, metastases, astrocytoma

There were several limitations in our present study. First, because of the rarity of CPTs, the sample size was small, especially for atypical CPP and CPC. However, this study constitutes the largest series of CPTs diagnosed according to the new classification scheme. In addition, no CT images were available to show the various intratumoural components. The final limitation was the relatively short follow-up period (mean 22.9 months) to monitor recurrence and metastases, although the estimated progression-free remission from diagnosis is approximately 48 months.[29] Further studies of the management and prognosis of CPTs will be needed.


In conclusion, MR is an effective tool not only to confirm the exact location of CPTs but also to provide better discrimination of parenchyma invasion and tumour vascularity. CPTs should be included in the differential diagnosis of a large intraventricular tumour that has multiple tortuous intratumoural areas of vascular flow voids. A mixture of nodular or patchy areas of inhomogeneous isointensity to slight hyperintensity on T2WI is a helpful feature for diagnosis of CPTs. A thin isointense capsule is suggestive of the benign subtype, while extensive peritumoural oedema or necrosis may predict aggressive CPTs. However, the variable degree and patterns of enhancement are rather non-specific. Characteristic signs can be accurately visualised overall by MR; thus, MR may prove helpful in distinguishing CPTs from the more common primary brain tumours and is useful for surgeons to determine the optimal treatment protocols of this highly vascular tumour.


The authors would like to thank Karki Bivek, MD, for his expert advice in preparing the manuscript.