The diagnostic radiographic approach to pineal tumors has changed over the past 30 years. Skull radiography, once an important study for detecting possible pineal neoplasm, has fallen into disuse because of its low sensitivity in detecting tumors. Unless the neoplasm is calcified, or unless the calcification in the pineal gland is displaced by a contiguous tumor or hydrocephalus, or unless changes of increased intracranial pressure are reflected in the sella, sutures, or calvarium, the plain x-ray film of the skull is not informative. Carotid and vertebral arteriography have been used in the past to identify the arterial blood supply to the pineal gland, venous drainage, and vascular displacement due to mass effect and obstructive hydrocephalus. However, at present, plain and contrast-enhanced computed tomography (CT) scans, with or without reconstruction (sagittal and coronal), are sufficient to make the diagnosis of a mass in the region of the pineal gland. In the last 20 years, magnetic resonance imaging (MRI) has proved to be the most informative imaging procedure for determining the anatomical localization of the mass, its relationship to adjacent vascular structures, and its spread by either direct invasion or subarachnoid dissemination. Both CT and MRI, while diagnostically highly accurate in determining the presence of a pineal region mass, lack specificity as to the histological nature of the tumor. Some relative specificity can be gained by paying attention to the density of the tumor on CT, the enhancement of the tumor on CT and MRI, and the intensity of the tumor on MRI. New in the diagnostic armamentarium is magnetic resonance spectroscopy. The role of magnetic resonance spectroscopy in the evaluation of pineal region tumors is, at present, relatively unknown but promising. Cerebral angiography retains a role in the evaluation of pineal region masses only in defining the arterial anatomy and venous relationships, information that may be useful when surgical resection is contemplated. Occasionally, characteristics of blood supply, such as dural tentorial blood supply (tentorial meningioma) may indicate the likely histological nature of the lesion.
A normal pineal gland measures 5 to 9 mm in length, 3 to 5 mm in height and up to 6 mm in width. The calcified pineal gland that is larger than 1 cm in any dimension should be looked upon with suspicion. Not only the size of the pineal calcification and its location are important but also the age at which the calcification appears. In the literature on the evaluation of skull roentgenograms it had generally been held that calcification of the pineal gland under the age of 10 was abnormal. However, Schey found that physiologic calcification could be seen in persons as young as 6 years. The overall incidence of calcification in childhood has been given by Willich et al. as 0.83 percent and by Peterson and Kieffer as 5.1 percent.
It is clear from the literature on pineal calcification detected on skull roentgenograms that the incidence varies according to the age of the population examined, the quality of the examination, and the genetic makeup of the population. The overall incidence of pineal calcification is around 13 percent.
Another factor in the incidence of skull radiographic evidence of pineal calcification is genetics. There is a body of literature that gives different rates for pineal calcification, according to the country of origin. Thus the incidence for Nigerians has been reported as 5 percent, for Japanese as 9.9 percent, for Fijians as 15 percent and for Indians as 19 to 24 percent. In contrast to these reports are the similar incidences found for calcification per decade of life by Adeloye and Felson in the American population and by Bhatti and Khan in the Pakistani population. For the first decade of life the incidence was zero in both series, for the second decade it was between 1.5 and 2.3 percent and for the third decade it was 10.5 percent. The incidence of pineal calcifications in both Americans and Pakistanis rises to 30 percent as the population ages into the middle and later years.
Roentgenograms of the skull may be of diagnostic value in the patient with a pineal tumor if (1) the tumor produces calcification that is visible on the skull films; (2) the pineal tumor produces obstructive hydrocephalus with increased intracranial pressure that demineralises the floor of the sella or separates the sutures or the dilated third ventricle amputates the dorsum sellae; or (3) the obstructive hydrocephalus causes inferior and posterior displacement of the pineal calcification(s).
The frequency of positive plain skull films in a series of pineal tumors varies. Abay et al., in a series of 24 patients with pineal tumors, found that in 25 percent the skull roentgenograms were abnormal on the basis of pineal tumor calcification, pineal calcification position, or evidence of increased intracranial pressure. Lin et al. reported a series of 32 pineal neoplasms, with calcifications seen in the region of the pineal in 75 percent. Of those with pineal calcification, the pineal gland was abnormal in size in 21 percent and in an abnormal position in 67 percent. The presence of the pineal calcification was abnormal in four of five patients who were 5 years of age or under (80 percent). The histologic classification of the tumors in the patients with abnormally large pineal calcification was teratoma in two, atypical teratoma in one, and pineoblastoma in two.
Factors that affect the detection of pineal calcification by CT are the thickness of the CT section, the size of the gland, the portion that is calcified, and the density of the calcification. Denser calcifications, thinner sections, and greater calcific portions make identification easier. It has been demonstrated that 8-mm-thick CT sections are eight times more sensitive than skull roentgenograms in the detection of 3-mm calcifications and 22 times more sensitive for 10-mm calcifications.
The youngest patient with a normally calcified pineal gland on CT was age 6.5 years in the series of Zimmerman and Bilaniuk. In this series, from ages 8 to 14 years the incidence of pineal calcification ranged between 8 and 11 percent. At age 15 the incidence rose to 30 percent, and at age 17, to 40 percent. Thus, the incidence of pineal calcification as detected by CT shows an increase that coincides with the onset of puberty. The presence of a small pineal calcification on CT, in and of itself, from age 6.5, upward is not evidence of a pineal neoplasm. Calcification under age 6 should be looked upon with suspicion.
CT has become an important diagnostic test for demonstrating the presence or absence of a pineal tumor. In the series of Abay et al. eight of nine CT studies were positive for pineal tumors. In one instance the CT was thought to be normal. In this false-negative study done on a first-generation CT scanner, the image quality was not ideal. In the first 44 pineal neoplasms diagnosed in the author's department since the advent of CT, there were two instances (4.5 percent) in which the pineal tumor was not appreciated initially. In one case. a 12-year-old girl presented with abnormal contrast enhancement of the subarachnoid space and a normal pineal region. Several subsequent CT examinations showed the same findings. Eight months after the onset of her headaches, a high-resolution. thin-section CT examination revealed a small pineal tumor. Biopsy revealed a pineoblastoma (primitive neuroectodermal tumor) and examination of the subarachnoid space revealed evidence of tumor seeding. The other patient had previously been treated for bilateral retinoblastomas. He was found to have an "incidental" CT finding of a pineal calcification at age 2 months. The patient was followed for a year, during which time a soft tissue mass grew around the calcification, producing obstructive hydrocephalus. Biopsy of this mass revealed a pineoblastoma. The first case represents a false-negative CT examination due to the small size of the lesion, while the second case represents a failure to appreciate the significance of a too-early-appearing pineal calcification.
Thirty-one patients with pineal, parapineal or histologically related tumors were reported by Zimmerman et al. The CT characteristics in that series allowed differentiation between benign (germinal) tumors, such as teratomas and epidermoids from malignant germinal tumors, such as the germinoma and embryonal cell carcinoma. Primary pineal tumors (pineoblastoma, pineocytoma) could not be differentiated from malignant germinal tumors on the basis of CT criteria alone. Germinomas appeared as soft tissue masses of slightly greater density than normal brain tissue. Calcification was not a feature of the tumor matrix in germinomas. Frequently the germinoma surrounded a centrally placed normal-appearing pineal calcification. In the smaller tumors, the germinoma was well defined and did not appear to invade the surrounding brain or subarachnoid spaces. Uniform contrast enhancement was the rule. With larger germinomas the margins became poorly defined and infiltration into the adjacent brain parenchyma and subarachnoid spaces became more common. Cystic changes within a non-operated tumor were unusual. Embryonal cell carcinomas had soft tissue densities similar to those of the germinomas and they commonly contained tumor calcification; but in contrast to the germinomas, the embryonal cell carcinomas more often showed cystic areas. The benign teratomatous tumors also showed cystic areas. In addition, the benign teratomatous tumors showed evidence of tissue derived from all three germinal layers, such as calcification, ossification, teeth, fat and soft tissue densities.
Epidermoid tumors of the pineal region have a density in the range of cerebrospinal fluid (CSF) and do not enhance with contrast administration. Thus, they may be confused with an arachnoid cyst or encysted ventricular structures.
Both the pineocytomas and pineoblastomas show an isodense to hyperdense tumor matrix, contrast enhancement and a tendency toward parenchymal calcifications within the tumor matrix. Two patients with pineoblastomas, when followed by sequential CT examinations prior to treatment, showed rapid tumor growth. All but one of the pineoblastomas presented within the first 12 years of life. The case in which the pineoblastoma presented later at age 40, occurred in the mother of the child who had presented at age 12 with a pineoblastoma. The increased incidence of pineoblastomas in patients with congenital bilateral retinoblastomas and the occurrence of pineoblastoma in mother and daughter raise the interesting question of a genetic predisposition in at least some patients.
Astrocytomas that arise within the pineal gland or adjacent to it will expand the gland, invade it or displace it. Since a glial stroma supports the pineocytes and is an integral part of the gland, some astrocytomas, presumably a small proportion of those involving the gland, arise directly from the pineal gland. Most often these tumors are of decreased density relative to brain parenchyma. The contrast enhancement that occurs is usually inhomogeneous.
Calcification occurs infrequently in these tumors, but when it does, its nature and position may make the differential diagnosis between astrocytoma of adjacent structures and non-astrocytic pineal tumor difficult. Because of their location, posterior hypothalamic astrocytomas and astrocytomas of the tectum of the mesencephalon may also be difficult to differentiate from primary pineal tumors on transverse section CT. They frequently abut on the cerebral aqueduct or on the posterior third ventricle, producing obstructive hydrocephalus similar to that produced by a primary pineal neoplasm. To differentiate these tumors by location, careful attention needs to be paid to the size and shape of the tectal region: the radiologist needs to look for forward displacement of the calcified pineal gland by an upper brain stem mass. Sagittally reconstructed sections may be of particular advantage in this situation, as may be the use of a subarachnoid non-ionic water soluble contrast agent at the time of CT sectioning.
In interpreting the CT findings, attention should be paid to the patient's age and sex. Teratomatous tumors of the pineal region occur almost exclusively in males. Germinomas occur in both males and females, most frequently in the second and third decades. Embryonal cell carcinoma occurs most often in the male in the second decade. Both germinomas and embryonal cell carcinomas are radiosensitive, but only the germinoma is radiocurable. After a period of regression, embryonal cell carcinoma tends to recur promptly. Sex distribution in the pineocytoma-pineoblastoma group seems equal. In a female patient with a calcified tumor in the pineal region, the most likely diagnosis is a tumor of primary pineal origin.
The major blood supply to the pineal gland arises from the posterior medial choroidal artery, a branch of the posterior cerebral artery. This vessel arises from the interpeduncular segment of the posterior cerebral artery. extends superiorly through the cistern of the lamina tecti and comes to lie on the lateral surface of the pineal gland. The artery continues on past the pineal to supply blood to the choroid plexus in the roof of the third ventricle. The pineal gland is drained by veins that originate from its superior and inferior surfaces. These veins drain into either the internal cerebral vein or the great vein of Galen. Superimposed upon the pineal vein anatomy on the lateral vertebral angiogram are the thalamic veins. It is difficult to differentiate pineal from thalamic veins. The internal cerebral vein lies in the roof of the third ventricle and then extends posteriorly through the velum interpositum into the cistern of the lamina tecti, where it joins the opposite internal cerebral vein and, with both basilar veins of Rosenthal, forming the vein of Galen. The vein of Galen lies just beneath the splenium of the corpus callosum and extends posteriorly to join the inferior sagittal sinus to form the straight sinus, which lies at the point of insertion of the falx cerebri onto the tentorium.
In the pathologic situation, a number of adjacent arterial and venous structures may be displaced or deformed or may provide an abnormal source of blood supply or venous drainage. How often these findings are appreciated angiographically is uncertain. It is probable that the degree to which these findings are appreciated depends upon the quality of the examination. A high-quality examination requires magnification angiography with subtraction, following the delivery of an adequate volume of contrast agent to the tumor bed in a patient who is able to cooperate. Abay et al. in a series of 12 pineal tumors studied by angiography, reported the presence of tumor vascularity or stain in four (33 percent).
More important than the presence of tumor stain is the recognition of the displacement of adjacent arteries and veins. Vascular displacements are dependent upon the direction of growth of the neoplasm and upon enlargement of the ventricular system resulting from obstruction of the posterior third ventricle or aqueduct. When the pineal gland is enlarged, the posterior medial choroidal artery is displaced posteriorly and laterally, and the superior and inferior pineal veins become more separated. The internal cerebral vein, which lies just above the pineal gland, is impinged on, elevated, and stretched. If the mass effect is sufficient, an angular deformity occurs at the point of juncture between the internal cerebral vein and the vein of Galen. If the tumor is large and extends posterolaterally, the vein of Rosenthal can also be deformed. As the tumor extends back into the cistern of the lamina tecti, down onto the tectum of the mesencephalon, and against the anterior-superior vermis, other changes occur. The superior cerebellar arteries are spread apart within the lamina tecti cistern (shown best with a Towne projection of the vertebral arteriogram); the precentral cerebellar veins are flattened and displaced from front to back (seen on the lateral projection of the vertebral angiogram); and the superior cerebellar artery branches, as they go over the anterior superior vermis, are flattened and deformed from front to back (seen on the lateral projection of the vertebral arteriogram). If the tumor extends forward (inferiorly), into the floor of the third ventricle, or as the third ventricle dilates from obstructive hydrocephalus, the thalamic perforating branches that arise from the proximal posterior cerebral arteries and posterior communicating arteries are displaced and stretched (seen on the lateral projection of the vertebral arteriogram). With obstructive hydrocephalus, the posterior lateral choroidal arteries (branches of the posterior cerebral artery that supply the choroid plexus of the lateral ventricles) are stretched. In the rare instance of a parapineal tentorial meningioma, selective injection of the internal carotid artery may show its dural blood supply, which is derived from the intracavernous portion of the internal carotid artery through the tentorial branch of the meningohypophyseal trunk. Both internal carotid arteriography and vertebral arteriography are of use in the differential diagnosis of pineal neoplasms when the lesion proves to be a vascular anomaly (such as a vein of Galen aneurysm) that stimulates a pineal tumor on CT examination.
Magnetic Resonance Imaging
Since 1983. MRI has come to play a major role in the pre- and postoperative evaluation of patients with pineal region masses. Superior anatomic localization is the first and foremost reason. MRI can show direct sagittal images of the pineal region, demonstrating the gland's relationship to the tectum of the midbrain, corpus callosum and posterior third ventricle. Coronal and axial images complement the sagittal views.
The combination of T1-weighted, proton density, and T2weighted images allows detection of abnormal alterations in signal intensity within the pineal gland and adjacent structures, such as the thalamus or brain stem. Tumors or disease processes characterized by an abnormal increase in interstitial water content within the lesion appear as regions of increased signal intensity on proton density images. CSF and fluidfilled cystic spaces appear high in signal intensity on T2-weighted images. Thus, the frequently occurring pineal cysts (a normal anatomic finding) are shown by a combination of T1 - and T2weighted images. On T1-weighted images, the cyst is hypointense and on proton density and T2-weighted images, bright.
The normal pineal gland tissue enhances with gadolinium on MRI. This is because the pineal gland lacks a blood-brain barrier. Thus, contrast enhancement within the pineal gland, in and of itself, does not denote abnormality. Mamourian and Towfighi used MRI to obtain images from six patients with pineal cysts measuring between 7 and 15 mm in size and found that with immediate imaging, enhancement initially showed a rim-like margin, but with delayed imaging (60 to 90 minutes) the cysts also enhanced because of diffusion of the contrast material into the cyst. Tamaki et al. evaluated 32 cases of pineal cysts and found that they did not enlarge on follow-up studies over the next 3 months to 4 years. None of the patients in either series was symptomatic secondary to the pineal cyst. It is thought that pineal cysts may arise from incomplete fusion of the third ventricular diverticulum that gives rise to the pineal gland. However, pineal cysts with both glial and ependymal linings have been found.
Calcification is best seen by CT and poorly seen by MRI. Calcification may be seen as a focal hypointensity, when the calcification occupies a sufficient portion of the volume of the slice. It is possible to have a pineal neoplasm that is not larger than the normal-size pineal gland but is identifiable on CT because of the presence of calcification too early in life. Such is the case in trilateral retinoblastoma. Under these circumstances it is possible to have the tumor not seen on MRI because the calcification cannot be visualized and the pineal gland is not increased in size and the enhancement of the pineal gland is considered a normal phenomenon.
MRI shows flowing blood as a hypointense flow void within the lumen of the vessel. This is true for both arteries and veins. Thus, on the routine T1-weighted, proton density and T2weighted images, the internal cerebral veins, vein of Galen and straight sinus can be identified clearly, along with their relationship to any pineal mass. Should the mass be a vascular malformation, then the hypointense flow voids that make up the mass can be identified as such and characterized as a vein of Galen malformation.
Despite the superior anatomic demonstration of a pineal mass by MRI it is not always possible to determine the site of origin. Thus, sometimes it is not possible to determine whether a mass in this region has arisen from the tectal plate, pineal gland or adjacent thalamus.
In the demonstration of dissemination of tumor into the subarachnoid pathways, gadolinium-enhanced MRI of the brain and spinal canal has proved superior to contrastenhanced CT and/or myelography with a water-soluble agent. Thus, the method of choice for determining the presence or absence of disseminated tumor is gadolinium-enhanced MRI of the brain and spine. This should be done before surgery in order to avoid confusion with postoperative blood products in the form of methemoglobin, which can be bright on T1-weighted images. MRI is also superior to CT in demonstration of blood products, whether acute, subacute or chronic. This is important in the case of occult vascular malformations, such as those arising in the midbrain, thalamus, splenium or corpus callosum, masses that may mimic pineal region tumors. In these instances. the pattern of signal intensity changes on T1-weighted, proton density and T1-weighted images may help characterize the presence of blood products, giving a pattern that suggests the presence of a vascular anomaly.
With teratoid tumors, the presence of fat produces increased signal intensity on T1-weighted images. That this is fat and not methemoglobin can be verified by the use of a fatsuppression pulse sequence, which will turn the high signal of fat to a low one but will leave the high signal of methemoglobin unaffected.
Germinomas on MRI appear as masses hypo- to isointense to gray matter on T1-weighted images. On proton density images they are often of slightly increased signal intensity, whereas on T1-weighted images they are most often iso- to hypointense. The reason for this decrease in signal intensity on long time to repetition (TR) images seems to be related to their dense cellularity. This signal intensity change is not unique to germinomas but occurs in lymphomas and primitive neuroectodermal tumors as well. The pineoblastoma is a primitive neuroectodermal tumor and has a similar signal intensity change on T1-weighted images. Germinomas enhance intensely. They are radiation-responsive and often disappear on imaging following the initial 3000 rad. Gadolinium-enhanced T1weighted images of the brain and spinal canal are used pre- and postoperatively to evaluate for disseminated disease. Choriocarcinomas and embryonal cell carcinomas arising in the pineal region have a more variable signal intensity on MRI. This is due to the frequent presence of haemorrhage in the tumor. Haemorrhage can have a variety of signal intensity appearances, including areas of hypo-, iso- or hyperintensity on T1-weighted, proton density, and T2-weighted images, depending upon the chemical state of the blood (oxy- or de oxyhemoglobin, intra- or extracellular methemoglobin, or hemosiderin). Teratomas of the pineal region often contain fat, which can be seen as a zone of increased signal intensity on T1-weighted images. Fat decreases in signal intensity on long TR images as the time to echo (TE) increases, and it disappears on fatsuppressed sequences. Pineoblastomas are a form of primitive neuroectodermal tumor. Calcification, if present in these tumors, is seen poorly or not at all on MRI. On T1weighted images they are hypo- to isointense masses; on proton density images the masses are of slightly increased signal intensity; and on T2-weighted images they are hypointense. They enhance strongly with contrast medium. The pineocytoma has a signal intensity pattern somewhat different from that of the pineoblastoma. On T2-weighted images pineocytomas are more often somewhat increased in signal intensity. Again, contrast enhancement is usually present. Astrocytomas arise from adjacent structures, such as the tectum of the midbrain, thalamus, and splenium of the corpus callosum, and intrinsically from within the pineal gland. These tumors are usually of low signal intensity on T1-weighted images and of increased signal intensity on proton density and T2-weighted images. Enhancement is variable and may or may not be present.
Arteriovenous malformations (AVMs) and fistulae within the thalamus and midbrain drain into the adjacent venous structures, such as the vein of Galen and straight sinus (the vein of Galen malformation), giving rise to their enlargement. These high-flow vascular structures, both arterial and venous, become enlarged hypointense flow voids on T1-weighted, proton density, and T2weighted images. Gadolinium can produce some enhancement within portions of the nidus of the AVM. Magnetic resonance angiography serves a role in anatomically delineating the feeding arteries, the nidus, and the draining veins.
Magnetic Resonance Angiography and Magnetic Resonance Spectroscopy
Within the last 14 years, magnetic resonance angiography has come into its own as a diagnostic technique. Two methods, time-of-flight and phase-contrast, have been used to produce images of flowing blood within vessels. Resolution remains a problem but is adequate to show major feeding arteries, the nidus of an AVM and draining veins. This is useful in the region of the pineal gland when there is a vein of Galen malformation. It can also be useful in giving a clear anatomical picture of the configuration of the internal cerebral veins, vein of Galen and straight sinus when a surgical approach is being contemplated for a solid tumor that is displacing or encasing these structures.
Proton magnetic resonance spectroscopy in the last 13 years has begun to come into its own as a diagnostic technique. Singlevoxel spectroscopy, using a 2 x 2 x 2 cm voxel size, can show the levels of choline, phosphocreatine, creatine, N-acetylaspartate (NAA), and lactate within the region studied. Preliminary work in paediatric brain tumors has shown that elevation of choline, a cell membrane metabolite, is increased to a larger extent in malignant tumors than in benign tumors. By calculating ratios of choline to NAA, an index of the tumour's relative malignancy can be determined.