Introduction	Choose Top of pageABSTRACTIntroduction <<Materials and MethodsResultsDiscussionReferencesCITING ARTICLES

Pineal cysts are a common MRI finding, seen in between 1.4% and 4.3% of random brain MRI studies [1, 2] and in 23% of MRI studies performed on healthy volunteers [3]. Pineal cysts are found in up to 40% of cadavers at autopsy [4, 5]. Considerable data exist to support the belief that these typical cystic lesions are truly incidental findings and are stable over time [6-9]. More complex cystic pineal lesions pose a problem for neuroradiologists, however. Pineal neoplasms—specifically, pineocytomas or germinomas—have been reported to appear similar to pineal cysts [10-12]. Studies have also shown that there are cystic lesions of the pineal region that probably are normal variants but that show complex imaging features [13, 14]. Concern that any pineal lesion showing more complex imaging features may be neoplastic has led to the recommendation that suspicious lesions should be followed over time to assess stability [15, 16], and this practice pattern is common at many institutions. Finally, large pineal cysts causing hydrocephalus and neurologic symptoms have been reported [17]. It is generally accepted that clinically significant pineal lesions can be discriminated from benign lesions by evaluating lesion stability over time. No reports to date, however, have investigated the yield of follow-up MRI in evaluating pineal lesions deemed suspicious by the reading radiologist. We conducted a retrospective study over 10 years to determine the natural history of incidentally discovered pineal region cysts that were followed over time specifically because of imaging features found concerning to the neuroradiologist at the initial reading.

Materials and Methods	Choose Top of pageABSTRACTIntroductionMaterials and Methods <<ResultsDiscussionReferencesCITING ARTICLES

Patient records yielded by a computerized search of radiology reports dated from 1998 to 2007 were reviewed. We retrospectively identified all MRI reports over that 10-year interval describing pineal cystic lesions that the interpreting neuroradiologist found concerning enough to recommend follow-up imaging. Fifty-four such cases were identified, 26 of which had follow-up MRI at our institution 6 months or more after the initial examination. Cases were excluded if there was a previous or interval surgery, pineal biopsy, or known pineal tumor (three cases). Patients were initially scanned for a wide variety of reasons (Table 1). No cases were recommended for follow-up imaging in the setting of emergent symptoms referable to the pineal region such as hydrocephalus or Parinaud syndrome. Follow-up imaging was not recommended in patients with known malignancy or immunosuppressed state, and none of the patients showed additional lesions or invasive disease.

View Larger Version

TABLE 1 : Clinical and Imaging Findings of 26 Patients With Pineal Lesions Considered Suspicious and Recommended for Imaging Follow-Up With Follow-Up of at Least 6 Months

View larger version (200K)

Fig. 1A —13-year-old girl (case 17 in Table 1) who presented with headache and syncope. Axial T2-weighted MR image of brain reveals T2-hyperintense lesion with internal septations. This lesion was initially reported as probable neoplasm.

View larger version (218K)

Fig. 1B —13-year-old girl (case 17 in Table 1) who presented with headache and syncope. Axial T2-weighted MR image obtained 23 months after A shows lesion to be stable.

View larger version (111K)

Fig. 2A —21-year-old woman (case 16 in Table 1) who presented with hemisensory deficit. Image from initial examination is not shown because images obtained before October 2005 were deleted from archive. Sagittal enhanced T1-weighted MR image obtained 14 months after initial examination shows 17-mm pineal lesion with thick capsule of susceptibility artifact, likely representing peripheral calcification. Reading radiologist thought this finding was concerning in 21-year-old patient.

View larger version (112K)

Fig. 2B —21-year-old woman (case 16 in Table 1) who presented with hemisensory deficit. Image from initial examination is not shown because images obtained before October 2005 were deleted from archive. Sagittal enhanced T1-weighted MR image obtained 24 months after A shows no change. This lesion was followed for 38 months.

Interpretations over the 10-year time interval of the study were performed by seven full-time neuroradiologists. Except for a senior member of the department with 35 years of full-time neuroradiology experience, all had a certificate of added qualification (CAQ) in neuroradiology at the time of the study. Retrospective review of cases was performed by the primary author, who was in the second year of a neuroradiology fellowship at the time, and the senior author who is CAQ-certified in neuroradiology.

Lesions were cataloged according to the description found in the initial radiology report that suggested follow-up imaging. For the purposes of this study, simple cysts met the criteria outlined by Barboriak et al. [9]: first, a round or ovoid area of signal abnormality centered in the pineal recess; second, that was hypointense to white matter and isointense to CSF on T1-weighted images and isointense to slightly hyperintense to CSF on T2-weighted images; third, that was internally homogeneous on T2-weighted images; and, fourth, with a uniform cyst wall ≤ 2 mm in thickness. The cyst wall can show thin homogeneous contrast enhancement. We further categorized simple cysts by size as ≤ 1 cm or > 1 cm.

Because reports did not always include a description sufficient to permit classification of the lesion into the scheme proposed, a retrospective review of images was necessary. A few discrepancies (n = 3) were identified; in those cases, the lesions were reclassified according to retrospective review as discussed in the Results section. We believe that the discrepancies were most likely due to an evolving classification system for pineal lesions [9] rather than interobserver variability. The reclassifications had no impact on the final conclusions of this study.

Atypical lesions were lesions described as showing a multiloculated or septated internal structure or that showed thick cyst walls or enhancement of the cyst wall measuring > 2 mm in thickness. Also included as atypical were cysts showing abnormal hyperintense signal on FLAIR images or thick susceptibility artifact, consistent with hemosiderin or calcification, on gradient-echo sequences. The latter finding is generally considered more concerning in younger patients (Figs. 2A, and 2B, case 16). As a subtype of atypical lesions we included probable mass if the initial report clearly emphasized a probable mass lesion or neoplasm as the primary consideration. These cases showed abnormal signal characteristics on short- or long-TR scans or showed solid enhancement.

The MRI equipment and protocols varied across the time of the study, and most comparable sequences were compared in the equivalent scan plane. Most studies were performed on 1.5-T magnets. Three studies after 2006 were performed on a 3-T magnet. Two studies were performed on a 1-T open magnet (Table 1). For T1-weighted spin-echo images, TRs ranged from 500 to 550 milliseconds and TEs from 8 to 15 milliseconds. T2-weighted images were generally acquired with fast relaxation fast spin-echo (FSE) technique with TRs ranging from 4,800 to 6,500 milliseconds and TEs from 90 to 110 milliseconds. For conventional T1 and fast relaxation FSE T2 sequences, the slice thickness was constant at 5 mm with 2- to 2.5-mm interslice gaps. All studies included a FLAIR sequence. All initial studies minimally included a gadolinium-enhanced axial sequence. Follow-up studies generally included a three-plane gadolinium-enhanced series. Gadolinium-enhanced follow-up studies were obtained either at a 5-mm slice width or as a single sagittal acquisition at a 1-mm slice width with multiplanar reformations into axial and coronal data sets. For these studies, the TR/TE was 8/3.7. All patients underwent gadolinium-enhanced MRI at the initial examination and on follow-up scanning. Images from all available examinations were reviewed.

View larger version (180K)

Fig. 3A —25-year-old woman (case 21 in Table 1) who presented with confusion and altered mental status. Image from initial examination is not shown because images obtained before July 2004 were deleted from archive. Sagittal enhanced T1-weighted MR image of brain obtained 63 months after initial examination shows simple pineal cyst with maximum dimension of 20.0 mm. Note pituitary adenoma.

View larger version (166K)

Fig. 3B —25-year-old woman (case 21 in Table 1) who presented with confusion and altered mental status. Image from initial examination is not shown because images obtained before July 2004 were deleted from archive. Sagittal enhanced T1-weighted MR image obtained 2 years 10 months after A is without change in maximal cyst dimension. This lesion was reported as stable for more than 8 years.

View larger version (150K)

Fig. 4A —39-year-old woman (case 26 in Table 1) who presented with chronic and worsening headache. Sagittal T1-weighted MR image of brain shows 15-mm complex thick-walled lesion centered in pineal region. No gadolinium was given at initial imaging, which was performed at 3 T.

View larger version (167K)

Fig. 4B —39-year-old woman (case 26 in Table 1) who presented with chronic and worsening headache. Sagittal T1-weighted MR image obtained 9 months later shows no significant change in size or T1 character of pineal lesion.

View larger version (184K)

Fig. 4C —39-year-old woman (case 26 in Table 1) who presented with chronic and worsening headache. Gadolinium administered at follow-up shows thick, masslike enhancement. This lesion proved to be pineal cyst at pathology.

Lesion measurements were made using a calibrated PACS measuring tool. All lesions were carefully measured and evaluated at each time point of follow-up. The longest lesion axis was invariably seen in the sagittal plane and provided the measurement indicated in Table 1.

Results	Choose Top of pageABSTRACTIntroductionMaterials and MethodsResults <<DiscussionReferencesCITING ARTICLES

The study group was composed of 20 females and six males (Table 1). Patients ranged in age from 2 to 75 years. The average age of the males was 44.5 years (SD, 19.2 years) and of the females, 26 years (SD, 14.4 years). Cases were followed for an average of 31.4 months, with an SD of 22.5 months. The longest duration of follow-up was more than 8 years (97 months), and the shortest was 7 months.

Imaging studies and radiology reports were retrospectively reviewed. The 26 lesions followed here were categorized as follows: simple cyst ≤ 1 cm, n = 5; simple cyst > 1 cm, n = 8; complex cystic lesions, n = 7; or probable mass, n = 6. In three cases, the initial report described a simple cyst that on retrospective review did not meet the criteria outlined by Barboriak et al. [9], and so those cases were reclassified as complex cystic lesions.

We found that differences in MRI technique, including differences in patient positioning and slice selection, and subtle motion artifacts could provide up to a 2-mm margin of error between studies. In practice, volumetric measurements were not routinely performed; we found that such measurements are subject to a greater degree of measurement error than the simple measurement of the longest axis of the lesion made on sagittal images. In one case, a follow-up report described a lesion as having slightly decreased in size, although on reevaluation the reported change in size can be accounted for by differences in MRI technique. We found that the greatest contributor to measurement error was slice selection when using a protocol with a 5-mm slice width.

In summary, no report on lesion follow-up indicated that a single lesion had grown or had new more concerning imaging features, and retrospective review of the images was in agreement with this reporting (Figs. 1A, 1B, 2A, 2B, 3A, 3B, 4A, 4B, and 4C).

Discussion	Choose Top of pageABSTRACTIntroductionMaterials and MethodsResultsDiscussion <<ReferencesCITING ARTICLES

Barboriak et al. [9] performed a retrospective study to assess the natural history of incidentally noted pineal cysts when diagnosed after another primary brain pathology. That study was specifically designed to establish the natural history of typical cystic lesions, and the investigators excluded all cases of complex or atypical lesions from their data set. They retrospectively followed 32 patients for 6 months to 9 years, concluding that incidentally noted simple, or typical, cystic lesions of the pineal region were essentially stable over the time interval of the study. We wished to expand that investigation to evaluate the natural history of indeterminate cystic lesions of the pineal region. Our study design is unique in that we present follow-up data of a subset of cystic pineal lesions found incidentally but with imaging features sufficiently worrisome that long-term follow-up was recommended by the neuroradiologist. We wished to learn the pretest probability of this decision and to evaluate the imaging appearances of those lesions over time. Over the interval of 1998-2007 we identified 26 lesions that were followed up for 6 months or greater. All of these lesions were stable over the time interval of follow-up, more than 8 years. No lesions were found to enlarge or to show more aggressive features at follow-up. We conclude that the yield of follow-up imaging of pineal lesions is extremely low.

The literature suggests that the appearances of pineal cysts and pineal tumors overlap [10, 11, 13, 18], although a recently published article questions the existence of cystic pineocytomas [14]. With the belief that neoplasms should show growth over time, time interval follow-up is commonly used as a tool to discriminate between stable cysts and growing neoplasms or to discriminate between stable and growing cysts. In order for this technique to have merit, however, it is important to know the relative frequency of pineal neoplasms and their expected growth rate. The frequency of normal-variant pineal cystic lesions versus the rarity of pineocytomas and cystic glial lesions of the pineal region may make the pretest probability of positive findings on follow-up so low that the expense of the examination cannot be justified. Second, pineocytomas and low-grade glial tumors are thought to grow very slowly [19], so a single 6-month, 1-year, or even 2-year follow-up may not show a significant change. Similar arguments can be made for following simple cysts that are proposed to grow into symptomatic lesions with obstructive hydrocephalus or clinically significant mass effect on the tectal plate.

Our study shows stability of suspicious pin eal lesions over time interval follow-up. We cannot know whether our study included any stable neoplastic lesions, although the literature would argue that most concerning pineal lesions are pineal cysts. Fleege et al. [13] examined the histology of 19 pineal lesions, 14 of which were presumed to be pineal neoplasms preoperatively; they found that all 19 were cysts at histology. In their study, they noted that the imaging appearances of pineal cysts include complex cysts and cysts with fluid levels, calcification, hemorrhage, and enhancement. Perhaps the most concerning feature of pineal cysts is the appearance of solid enhancement, which is thought to be due to gadolinium diffusing into the cyst over time [20].

In our study, lesions were followed based on the initial reading radiologist's suspicion for tumor; thus, there is a level of subjectivity regarding the threshold for recommending follow-up imaging. We believe that our data set includes all of the most suspicious lesions seen at our institution over a 10-year period and that our results support the claim that no pineal lesions grew or were defined as more likely neoplastic on the basis of follow-up imaging.

Because clinical MRI is evolving with regard to both hardware and sequence protocols, retrospective studies invariably include studies performed on various MR machines and with varied sequence parameters, a problem that clinical radiologists face daily. Nevertheless, reliable clinical decisions are made when comparing such studies, and in this retrospective analysis, given the differences in techniques, no significant difference in the size or character of the pineal lesions was noted when comparing studies over time.

Our data set includes five patients younger than 16 years (cases 10, 13, 17, 18, and 20 in Table 1). The frequency of pineal tumors is thought to be higher in children [21], and pediatric cases may constitute a subpopulation for analysis. Although pineal tumors, in both children and adults, are occasionally diagnosed at this institution, such a tumor has not been identified through follow-up imaging.

Our institution is located in a rural area and is the only tertiary care referral center in the state. As such, our patient population tends to return for long-term follow-up, making this study possible. Our data set is relatively small; however, the lesions followed here are a subset of cystic lesions of the pineal region—worst-case scenarios for diagnostic imaging in the sense of being found incidentally but with concerning imaging characteristics. Larger data sets are needed to derive true statistics regarding pretest probabilities for follow-up decisions. Based on our experience, the yield of temporal follow-up imaging of concerning pineal lesions appears to be extremely low. With increasing interest in optimal allocation of health care resources, we suggest that incidentally identified pineal region cystic lesions, both typical and atypical, can be managed clinically.