Homogeneous cystic abnormalities of the pineal gland, called pineal cysts, are common incidental findings on brain MR imaging [1, 2]. Large cysts rarely cause symptoms of paralysis of upward gaze and headache because of mass effect on the tectum and hydrocephalus due to compression of the cerebral aqueduct [3]. It has been suggested that imaging follow-up may be helpful to document stability of size of incidental pineal cysts with atypical imaging features [4]. It has also been suggested that imaging follow-up should be performed on particularly large cysts (> 1 cm [5] or > 1.4 cm [6]). To study the natural history of asymptomatic pineal cysts, we retrospectively reviewed the clinical and imaging characteristics of 32 patients with pineal cysts examined with a 6-month or more interval between imaging.

A computerized search of radiology report texts identified 45 patients with cystic abnormalities of the pneal gland who had undergone at least two brain MR imaging studies separated by at least a 6-month interval between 1986 and 1998. Patients were included in the study if a cystic abnormality of the pineal gland was reported on any brain MR image. Patients were then excluded from the study if there was a history of tumor at or adjacent to the pineal gland (three patients), if review of the MR studies failed to show a typical pineal cyst (three patients), or if necessary images were missing or destroyed (seven patients). This elimination left 32 patients who formed the study group. For the purposes of this study, a typical pineal cyst was considered present on MR study if the following four criteria were met: first, a round or ovoid area of signal abnormality was seen centered in the pineal recess. Second, the abnormality was hypointense to white matter on T1-weighted images and isointense with cerebrospinal fluid on T2-weighted (long TR, long TE) images. Third, the abnormality was internally homogeneous on T2-weighted images. Fourth, no lobularity of the margin or nodular contrast enhancement was present. A rim of either hypointensity on T2-weighted images or contrast enhancement on T1-weighted images was allowable if the rim measured less than 2 mm thick.

The indications for MR imaging were obtained by review of computerized medical records. In four patients, the major indication for MR imaging in the report had been to follow up a known pineal cyst. Eighteen patients were scanned for follow-up of intracerebral neoplastic disease, which included pituitary adenoma in five patients, gliomas in six patients, medulloblastomas in four patients, metastases in two patients, and ependymoma in one patient. Two of the patients with intracerebral neoplastic disease carried a diagnosis of neurofibromatosis. Three patients were scanned to examine demyelinating disorder, two patients were HIV-positive (one with intracerebral lymphoma), two had a seizure disorder, one had Bell's palsy, one had acute lymphocytic leukemia, and one had a history of cerebral infarctions.

For each patient, two sets of cranial MR images of the brain were reviewed. The earliest available cranial MR images were used to examine the pineal region at the initial time point, and the most recent available cranial MR images were used to examine the region at the final time point. Because of the variability of MR protocols used for brain imaging during the 13-year period covered by this study, in general the pulse sequence parameters used at the initial time point were not identical to those used at the final time point. Images obtained at the initial and final time points were matched on the basis of both image contrast (contrast-enhanced T1-weighted, unenhanced T1-weighted, proton density-weighted, or T2-weighted image contrast) and scan plane (axial, coronal, or sagittal planes).

All MR imaging was performed on 1.5-T magnets. Proton density-weighted and T2-weighted images were acquired with a conventional spin-echo technique with TRs ranging from 2300 to 3000. TEs ranged from 30 to 40 for proton density-weighted images and from 70 to 100 for T2-weighted images Slice thickness ranged from 4 to 5 mm with 2- to 2.5-mm interslice gaps. For T1-weighted spin-echo images, TRs ranged from 500 to 700 and TEs ranged from 8 to 20. Slice thickness ranged from 3 to 5 mm with up to 2.5-mm interslice gaps.

In all 32 patients, available matching images allowed measurement of the pineal cyst in two orthogonal directions. In 23 patients, available matching images allowed measurement of the pineal cyst in three orthogonal directions at both time points.

A single reviewer used an 8-power photographer's loupe to measure linear dimensions of the pineal cyst (including cyst rim, if present) and of the distance scale on MR images to the nearest 0.1 mm. In general, the measurements obtained on films from the initial and final time points were performed at separate sessions, but on difficult cases, the initial and final images were measured side to side to ensure that comparable measurements were obtained. At a later time, the actual size of the pineal cyst was calculated with a spreadsheet program by dividing the size of the cyst on film by the size of the distance scale on film and multiplying by the actual size represented by the distance scale.

Twenty-five measurements of pineal cyst dimension chosen at random were replicated by a second reviewer blinded to the results of the initial reviewer to evaluate interobserver variation. The mean (±SD) difference between actual pineal cyst dimensions based on measurements obtained by the two reviewers was 0.02 ± 0.68 mm, with a range of -1.1-1.4 mm.

To compensate partially for differences in head position, the maximal linear dimension of the pineal cyst was recorded in each patient at both time points. In general, differences in slice selection and the presence of interslice gaps made precise matching of initial and final images impossible. Because of this discrepancy, changes in the maximal linear dimension of cysts less than 2 mm (actual distance) were arbitrarily considered not significant. Cyst volumes were calculated in 23 patients by multiplying measurements in three orthogonal directions.

Proton density-weighted images (available in 26 patients at the initial time point) and contrast-enhanced images (available in 22 patients at the initial time point) were examined for pineal cyst signal intensity and enhancement pattern. The amount of deformity of the tectum due to pineal cyst was graded on all patients on both initial and final images with a five-point scale. A grade of 0 signified no contact between the cyst and the tectum, 1 signified the cyst touching the tectum, 2 signified the cyst mildly deforming the tectum but not narrowing the aqueduct, 3 signified the cyst moderately deforming the tectum but not narrowing the aqueduct, and 4 signified the cyst deforming the tectum and narrowing the aqueduct. Lateral ventricular size was graded in all patients on both initial and final images as normal, minimally enlarged, mildly enlarged, moderately enlarged, or severely enlarged.

Radiology reports and electronic medical records were reviewed in each patient to determine the indication for MR imaging and to record any history of surgery to the pineal area, symptoms referable to the pineal gland such as paralysis of upward gaze or paroxysmal headache, or symptoms of hydrocephalus.

The time course of change in dimension of pineal cysts was evaluated on available brain MR images obtained between the time of initial and final scans in those patients who showed either pineal cyst development or pineal cyst involution during the study period.

The study group comprised 19 females and 13 males. The mean (± SD) age of patients was 22.9 ± 17.0 years, with a range of 0.7-67.4 years. Fifteen patients were less than 18 years old, and only one patient was more than 60 years old. No history of symptoms specifically referable to the pineal gland was recorded for any of the patients during the study period, and no surgical treatments for pineal cyst were performed.

The characteristics of the pineal cysts at the initial scan are described in Table 1. The maximal linear dimension of the pineal cyst on initial study was 10 mm or less in 15 patients (47%), between 10 mm and 15 mm in 11 patients (34%), between 15 mm and 20 mm in four patients (13%), and greater than 20 mm in two patients (6%). All cysts were homogeneous on proton density-weighted images (n = 26). The mean time of follow-up was 3.7 ± 2.6 years with a range of 0.5-9.1 years. Nine patients were followed up for 5 years or more.

In the 23 patients in whom matching measurements in three orthogonal directions were possible, no significant change was seen in the mean cyst volume at initial scan and at final scan (1.42 ± 1.45 cm³ versus 1.40 ± 1.41 cm³, p = 0.95). In an analysis of the entire study group, no significant difference was seen between the mean maximal linear dimension of pineal cysts measured at initial scan and at final scan (11.2 ± 4.4 mm versus 10.9 ± 4.6 mm, p = 0.64). No significant change in maximal cyst dimension (change of <2 mm) was found in 24 (75%) of 32 patients. Two cysts resolved completely on follow-up (Fig. 1A,1B,1C,1D,1E,1F), three cysts decreased by 2-4 mm, two cysts enlarged by 2-3 mm (Fig. 2A,2B), and one cyst formed and grew to 1.2 cm (Fig. 3A,3B,3C).

The time course of change in maximal cyst dimension was studied in the two patients whose cysts resolved completely and in the patient who developed a pineal cyst. An 8.8-mm cyst in a 9-month-old girl with a right optic nerve glioma remained unchanged in size through the age of 2 years but was not seen after the age of 3 years. A 10.6-mm cyst in a 7-year-old girl with neurofibromatosis and thalamic pilocytic astrocytoma remained unchanged through the age of 10 years, then decreased in size to 8.8 mm and 6.3 mm at the age of 12 years. No significant cyst was seen at the age of 13 years (Fig. 1A,1B,1C,1D,1E,1F). On follow-up scanning 1 year after his initial scan, a 5-year-old boy with disseminated medulloblastoma had developed an 11.6-mm cyst that remained stable in size through the final scan at the age of 8 years (Fig. 3A,3B,3C).

No significant correlation was seen between the duration of follow-up and the magnitude of change in either cyst volume (r = 0.16) or maximal linear dimension (r = 0.2) over the study period. No correlation between the cyst volume at the initial scan and change in cyst volume was found (r = -0.25). A trend toward an inverse relationship between maximal cyst dimension at the initial scan and change in maximal dimension was noted (r = -0.33, p = 0.06). Both cysts that increased by 2 mm or more in maximal dimension were less than 10 mm in dimension on the initial study.

There was no correlation between age at initial scan and change in maximal dimension (r = 0.12). As noted previously, the three patients in whom cysts either formed or involuted were children.

On initial study, 22 of 25 cysts were hyperintense to cerebrospinal fluid on proton density-weighted images. One cyst that was initially isointense to cerebrospinal fluid was hyperintense on the final scan. No nodular enhancement of pineal cyst was present on the final study (n = 31).

On the initial study, higher grades of tectal deformity were associated with a larger average maximal linear dimension of pineal cysts (p < 0.001). The grade of tectal deformity on the initial study was not associated with increasing maximal cyst dimension over the study period. In the patient who developed a 1.2-cm pineal cyst, the final study showed contact of the tectum with pineal cyst (Fig. 3A,3B,3C). In every other patient, there was no increase in grade of tectal deformity when initial and final studies were compared.

When the initial and final studies were compared, evidence of increasing ventricular size was seen in only two of 32 patients. The first of these patients was an HIV-positive 37-year-old man with central nervous system lymphoma, whose lateral ventricles, initially normal-sized, became minimally enlarged. The second was a 41-year-old woman with intracerebral metastases from renal cell carcinoma being treated with steroids and antiseizure medications, whose lateral ventricles, initially normal-sized, became mildly enlarged. In both patients, ventricular enlargement was ascribed to increasing atrophy rather than hydrocephalus. In two patients with pineal cysts greater than 2 cm in maximal dimnesion, the ventricles were unchanged in size and no more than mildly enlarged.

Cystic lesions of the pineal gland are found in between 21% and 41% of autopsy specimens [7] and in between 1.4% and 4.3% of cranial MR studies [1, 2] and are rarely related to clinical symptoms. These lesions may, however, become clinically important for two reasons: first, pineal cysts may enlarge over time (because of either increased cyst fluid or intracystic hemorrhage) and become symptomatic. Mass effect against the adjacent midbrain may lead to Parinaud's syndrome (paralysis of upward gaze, lid retraction, and abnormal pupillary reactions), whereas compression of the cerebral aqueduct may cause symptoms related to hydrocephalus [8]. Cases of sudden death from intracystic hemorrhage and acute hydrocephalus called “pineal apoplexy” have been reported [9, 10]. Second, benign-appearing cysts of the pineal gland may represent malignant or premalignant conditions. Rarely, pineocytomas may present as purely cystic tumors [11], or tumors may develop on the pineal cyst wall [4]. A recent study showed considerable overlap between the appearance of pineal cysts and pineocytomas on MR imaging, although a thin smooth cyst wall measuring no more than 2 mm thick was seen only in benign glial pineal cysts [12]. Given the frequency of pineal cysts on routine cranial MR studies, we believe that it would be impractical to follow up each incidentally discovered asymptomatic pineal cyst. For this reason, the recommendation has been made to follow up cysts with atypical imaging features [4].

A rational protocol for the follow-up of asymptomatic pineal cysts detected by MR imaging awaits the resolution of three problems. First, the imaging features of pineal cysts that should prompt follow-up imaging remain unclear. Although it has been recommended that particularly large pineal cysts (> 1 cm [5] or > 1.4 cm in diameter [6]) undergo follow-up imaging, no evidence has been presented to suggest that larger cysts are more likely to grow or become symptomatic. Second, the significance of small changes in pineal cyst size seen on follow-up MR imaging is unknown. In part, this lack of knowledge is because the natural history of incidentally discovered asymptomatic pineal cysts is incompletely understood. The finding that pineal cysts are relatively rare in young children [13] and that on average, these cysts are seen in adults less than 40 years old [2, 14, 15] suggests that pineal cysts may typically form in late childhood and later involute. Thus, an interval increase in cyst size could simply reflect the natural history of asymptomatic cyst development and may not presage the development of symptoms. Finally, appropriate intervals for imaging follow-up of atypical pineal cysts and criteria for cessation of cyst surveillance are undetermined.

Our purpose was to study the natural history of pineal cyst size changes and imaging characteristics on long-term MR imaging follow-up. We carefully selected cystic pineal abnormalities that did not show nodular enhancement along the cyst wall or inhomogeneity of cyst contents on the initial study and confirmed that no symptoms specifically referable to the pineal gland were present at the time of either the initial or final imaging. Our findings show that small changes in the size of pineal cysts are infrequently seen on MR imaging. When stability was defined as a change of 2 mm or less, 75% of pineal cysts were stable over the follow-up period, whereas small (between 2 and 3 mm) increases in maximal cyst dimension were only seen in two (6%) of 32 patients. Our finding of cyst involution in two patients was similar to that of Tamaki et al. [16], who showed involution in two of 32 patients. We also showed pineal cyst formation in one patient between the ages of 4 and 5 years.

Except in a single case, the signal characteristics of the pineal cysts on proton density-weighted images were unchanged. The relatively young average age of our patients (22.9 years) is in keeping with prior studies showing that cysts are on average seen in patients less than 40 years old [2, 14, 15]. This young average age may also be partially due to biases in the underlying population of patients who undergo long-term follow-up imaging for indications generally unrelated to pineal cyst.

These asymptomatic cysts were associated with various degrees of tectal deformity. As expected, higher grades of deformity were seen in larger cysts, but we found no evidence that cysts with higher grades of deformity were more likely to enlarge on follow-up study. Our finding that the size of lateral ventricles remained relatively unchanged was expected, given the range in size of pineal cysts found in our study. In a prior study, all patients with obstructive hydrocephalus had cysts greater than 2 cm in diameter [17], whereas in our study, only two patients with unchanged, no more than mildly enlarged, ventricles had cysts greater than this size threshold.

Prior studies have found that pineal cysts generally do not change in size. Tamaki et al. [16] showed no change of cyst size in 29 of 32 patients during a follow-up period of 3 months to 4 years, whereas Golzarian et al. [15] reported on 12 patients with pineal cysts that did not progress after 1 year or more of follow-up. These studies did not, however, specify how interval changes or lack thereof were determined or whether linear cyst dimension or cyst volume was used. In addition, few details about the duration of imaging follow-ups are provided. Bodensteiner et al. [18] found two pineal cysts in 27 healthy adults imaged with T1-weighted sequences using 3-mm contiguous slices; neither subject with pineal cyst showed a significant change in cyst volume when rescanned after a 6-month interval with the same protocol.

The finding that pineal cyst development, small changes in cyst size, and cyst involution can all be seen on serial MR imaging without the development of specific symptoms supports the suggestion that typical pineal cysts found incidentally on MR imaging may be followed on a clinical basis alone, rather than by imaging. Further, the finding that the length of follow-up was not correlated with the magnitude of size changes over the period of observation suggests that increasing the interval between follow-up MR scans may not increase the likelihood of finding a change in pineal cyst size. Although approximately half the pineal cysts in this study were more than 1 cm in maximal linear dimension, we found no evidence to suggest that these larger cysts were more likely to enlarge over the duration of the study period.

Several limitations of our study are apparent. First, most of our patients underwent serial MR imaging for indications unrelated to the presence of pineal cysts. Specifically, 20 of 32 patients underwent serial imaging for intracranial neoplastic disease. It is unclear whether the presence of intracranial neoplasms or treatments for neoplastic disease could alter the natural history of pineal cysts. One previous autopsy study [19] showed pathologic cystic changes of the pineal gland characterized by multiple slitlike cavities, gliosis, and large numbers of Rosenthal fibers in 31 of 275 patients with neoplastic disease. Interestingly, 24 of the 31 patients with cystic pineal changes carried the diagnosis of acute leukemia, a diagnosis only found in one of our patients with a stable pineal cyst. Second, changes in MR imaging protocols over time and the lack of high-resolution thin-section imaging in this patient series make reliable detection of small changes in cyst size difficult. Radiologists in clinical practice often face similar difficulties, and we expect that our results would be applicable to routine clinical practice with standard MR imaging protocols. Finally, the underlying pathophysiologic process that causes an asymptomatic pineal cyst to enlarge and become symptomatic is not understood and appears to occur infrequently. A much larger sample size than ours may be needed to determine whether any imaging characteristics of incidental pineal cysts are associated with an increased risk of subsequent enlargement or the development of symptoms.

In summary, the size of pineal cysts as a whole remained unchanged on serial MR studies, but in individual patients cysts were seen occasionally to either form or involute. Small increases in cyst size were uncommonly seen and were not associated with specific clinical symptoms. We found no evidence to support the recommendation that follow-up studies should be performed for pineal cysts measuring more than 1 cm. Our findings suggest that typical pineal cysts found incidentally on MR imaging may be best followed up on a clinical basis alone.

Address correspondence to D. P. Barboriak.