ORIGINAL ARTICLE 

Annals of Nuclear Medicine Vol. 7, No. 4, 257-263, 1993 

Reassessment of quantitative thallium-201 brain SPECT for miscellaneous brain tumors 

Shigeru KOSUDA,* Hirofumi FuJn,** Shigeki AoKI,** Kenzo SuzUKI,** Yoshiaki TANAKA,** Osamu NAKAMURA*** and Nobuyuki SHIDARA*** 

* Department of Radiology, National Defense Medical College * * Department of Radiology, Tokyo Metropolitan Komagome Hospital * * * Department of Neurosurgery, Tokyo Metropolitan Konlagome Hospital, Tokyo, Japan 

In order to reassess the value of quantitative thallium-201 brain SPECT in the differen-tiation of miscellaneous brain tumors, we studied a total of 89 patients-35 pre-operative patients suspected of having a brain tumor and 54 post-operative patients with a brain tumor. We came to the conclusion that quantitative Tl-201 brain SPECT was very useful in discriminating cerebral radiation necrosis from recurrent tumor, estimating residual tumor burden, and detecting tumor regrowth earlier in postoperative patients. In preoperative patients, however, T1-201 SPECT cannot be used effectively to differentiate glioma from other intracranial tumors, although intense uptake of T1-201 may provide evidence of glioblastoma or a hypervascular lesion. Key words 201T1 chloride bram tumor cerebral necrosrs radiotherapy, brain SPECT 

INTRODUCTION

THE DIFFERENTIATION Of cerebral necrosis after radio-therapy from recurrent glioma is a clinically chal-lenging problem. Although magnetic resonance imaging (MRD and computed tomography (CT) are excellent anatomic imaging methods for detecting mass lesions, both have limitations in discriminating between cerebral radiation necrosis and recurrent glioma,1-5 as well as in the grading of glioma.6 Positron emission tomography (PET) studies with fluorine-18-fluorodeoxyglucose (18F-FDG) and carbon-11-methyonine have been shown to assist in discriminating between cerebral radiation necrosis and tumor recurrence, but the limited availability and high cost of performing PET studies restrict clinical application.7-9 
It has been reported that thallium-201 (201Tl) is preferentially taken up by high-grade astrocytoma, but little or none is taken up by low-grade astro-cytoma or cerebral necrosis.10-15 Quantitative 201T1 single photon emission tomography (SPECT) may play a role in differentiating cerebral necrosis from tumor recurrence. However, the value of 201Tl SPECT remains uncertain and several questions are unanswered in the clinical setting. Firstly, a pre-liminary comparative study with 201T1 and 18F-FDG has suggested a disparity in the detection of recurrent glioma and the differentiation between cerebral necrosis and recurrence.16 Secondly, few series of 201T1 SPECT studies have been published contain-ing sufficient numbers of cases to prove the value of aolTl SPECT. Thirdly, it remains to be seen whether 201Tl SPECT will be of value in the preoperative differentiation of glioma from other miscellaneous brain tumors. 
The purpose of this study is to reassess the value of 201Tl SPECT in differentiating between glioma and other malignant or benign lesions before opera-tion and between cerebral radiation necrosis and recurrence of brain tumor in follow-up. 

MATERIALS AND METHODS 

A total of 89 patients-35 preoperative patients with suspected brain tumor and 54 postoperative patients with confirmed brain tumor-underwent 201T1 SPECT in order to assess preoperative findings, postoperative findings, and the implication of delayed imaging. These 89 patients were 48 males and 41 females, with an age range of 14 to 80 years (mean age, 56 years). The disease classification is shown in Table I . Histological confirmation was available in all brain tumors. Recurrent brain tumors in 26 patients and cerebral radiation necrosis in 6 patients were all confirmed by re-operation. The 20 patients with recurrent glioma were histologically classified as having glioblastoma (10), astrocytoma Grade 111 (9) and gemistocytic astrocytoma (1). Sixteen patients treated with trimodal therapy, consisting of surgery, radiotherapy and chemotherapy, had no evidence of residual tumors on concomitantly available CT and/or MRI with contrast medium. Among these 16, nine had histological glioblastomas, three anaplastic astrocytomas, two gemistocytic astrocytomas, one oligodendroglioma, and one Grade 11 astrocytoma. Seven patients died, 6 of recurrent tumor at an average of 19 months after the 201T1 study, and one of pneumonia at 2 months after the 201Tl study. Four patients are alive as of this writing, but one of these 4 developed recurrence at 39 months after the 201Tl study. Three patients were lost to follow-up and 2 patients were transferred to other hospitals. 
Non-neoplastic diseases, including cerebrovascular disease, abscess, encephalitis, and arteriovenous malformation (AVM), were diagnosed from the clinical course and by imaging procedures, such as CT. MRI, and angiography. A11 but one of the 54 patients in the postoperative group had a follow-up of at least one year after the initial 201Tl SPECT study in order to assess prognosis. 
All of the 89 patients had early SPECT images taken 10 minutes after iv injection of 1 1 1 MBq (3 mCi) of 201Tl. Of these, 16 patients also had delayed SPECT images taken 4 hours after the injection. All images were acquired over 32 minutes with continuous rotation over 360', by means of a dual head rotation camera (Toshiba 90A E1) with 20 mm spatial resolution in FWHM. We used a 10% symmetrical window at 71 KeV. Each reconstructed slice was two pixels thick (10.6 mm). A 64 x 64 matrix with a Butterworth filter and filtered back projection by means of Chesler's algorithm were used to reconstruct images in the transverse plane. A uniformity correction matrix and a center of rotation correction were applied to each image. Neither tissue attenuation correction nor scatter subtraction was processed. 
An operator-defined, rectangular region of inter-est (ROI) was drawn over a lesion on the slice showing the greatest activity. In creating a rectan-gular ROI, we tried to make a larger ROI in order to cover an entire lesion. Similarly, a ROI over the contralateral, presumab]y healthy, brain was created by performing a horizontal flip of the initially defined ROI. But a ROI was created in the normal area adjacent to a lesion when the lesion was located on the midline. Count ratios of a lesion to the normal brain (L/N) were calculated from the rectan-gular ROI for the quantitative analysis. Furthermore, count ratios of the lesion seen on the delayed image to that seen on the early image (D/E) were calculated. Ten postoperative patients with brain tumor under-went follow-up 201T1 imaging to monitor tumor recurrence. 

RESULTS 

1. Preoperative assessment Grading ofgliomas: Three patients with glioblas-toma had L/N ratios of 1.6, 2.8, and 6.4, respectively. In 8 patients with Grade ll astrocytoma, on the other hand, there was a subtle uptake to no uptake of 201Tl, making it impossible to create a ROI over the lesions. The ratio in these individuals was rated at 1.0 (Fig. l). 

Differentiation between glioma and other miscella-neous tumors: In 2/2 patients with meningioma and 7/8 (87.5~) with cerebral metastasis, L/N ratios were greater than 2.5. This was in contrast to the 2.5 or less in all patients with non-neoplastic diseases. The L/N ratio was 22.1 for hypervascular metastasis from thyroid cancer, 3.96 for hemangioblastoma, 2.51 for cavernous angioma, and 2.29 for malignant lymphoma. 

2. Postoperative assessment 
Differentiation between radiation necrosis and tumor recurrence: In 32 patients with recurrent or residual tumor, L/N ratios ranged from 1.7 to 12.6, with a mean of 4.1 5~2.18 on early images. All but one of the patients had L/N ratios greater than 2.5. Pathological diagnoses were glioblastoma (n=14), anaplastic astrocytoma (n=10), malignant menin-gioma (n=2), and cerebral metastases including adenocarcinoma (n=3), squamous cell carcinoma (n=1), renal cell carcinoma (n=1), and seminoma (n = 1 ). 
In 6 patients with cerebral radiation necrosis, however, L/N ratios were a]ways 2.5 or less on early images, with a mean of 2.06+-0.25 (p<0.001) (Fig. 2). These results confirmed the validity of 201Tl brain SPECT in differentiating cerebral radiation necrosis from tumor recurrence. 

Differentiation between residual tumor and post-therapy cerebra/ changes: All of the 6 patients with residual tumor after operation had L/N ratios of greater than 2.5, with a mean of 5.04-1.46. All but one the 16 patients who had no evidence of residual tumor after having undergone combined treatment with surgical debulking, radiotherapy and chemo-therapy had an L/N ratio of 2.5 or less, with a mean of 1.51+-0.58 (p<0.001) (Fig. 3). Among these 16 patients, eight had visibly increased uptake on early images. 

Follow-up studies in post-operative patients with brain tumor: In follow-up 201Tl SPECT studies of 10 post-operative patients, an L/N ratio greater than 2.5, as calculated on early images, correlated with either a residual or a recurrent tumor, but an L/N ratio of 2.5 or less is associated with an absence of residual tumor (Fig. 4). In two patients with recur-rent tumor, the L/N ratio decreased to 2.5 or less after treatment ; one of these 2 had pathologically proven cerebral radiation necrosis before developing tumor recurrence. A 28-year-old male patient with gemistocytic astrocytoma in whom the L/N ratio increased from 1.0 to 2.4 after treatment, as shown in Fig. 4, is still alive at 4 years after the 201T1 study. 3. Implication of delayed images Of 13 patients with residual or recurrent tumor, L/N ratios on the delayed images were less in 9 patients (69%) and slightly greater in the remaining 4 patients (13%). The three patients with marked washout of 201Tl had recurrent glioblastorna (Fig. 5). In a group of patients with cerebral radiation necrosis, there were little differences between the L/N ratios for early and delayed images (Fig. 5). Mean D/E ratios were 1.17+-0.20 for both tumor recurrence and residual tumor, and 1.17+-0.08 for cerebral radiation necrosis. For healthy brain paren-chyma (n=89), it was 1.32+-0.25. 

DISCUSSION 

The prognosis of high-grade glioma remains poor, in spite of recent advances in combined method treatment. High dose radiotherapy is an integral part of the treatment because of the limited ability to achieve total surgical resection of the lesion. How-ever, there is evidence to show that cerebral radiation necrosis is a frequent event in patients with high dose radiotherapy. The majority of radiation necrosis cases were depicted on initial CT scans as central low density areas with irregular peripheral enhancernent and as an increase in size with peripheral hypodense areas on follow-up scans. It is well known that cerebral necrosis and recurrent brain tumor are essentially indistinguishable by CT, MRI, and clinical findings.1-5 
201T1 SPECT is known to be useful in differen-tiating radiation necrosis from recurrent primary brain tumor. It has also been shown that 201Tl is preferentially taken up by viable tumor cells but not by necrotic tissues,n and that it may be subtly taken up by cerebrovascular disease.10,13 Black et al,12 reported that 18F-FDG PET and 201Tl scans closely correlated with each other in a group of patients with brain tumor, although the number of patients in their series was too small for definitive conclusions to be drawn. On the other hand, McKusick et al.16 reported in a preliminary series that 201Tl imaging was inferior to 18F-FDG PET imaging for the detection of recurrent primary brain tumor. 
Our quantitative 201Tl brain SPECT study in a large number of patients revealed that the 201Tl uptake index (L/N ratio) is useful in discriminating cerebral radiation necrosis from recurrent tumor. In addition, 201T1 was found to be taken up by residual tumor, a finding in agreement with those of earlier studies, confirming the validity of quanti-tative 201T1 brain SPECT in the postoperative patient population with cerebral tumor. There were differences between L/N ratios in residual tumor and post-therapy changes, making it possible to differentiate between them. The results of follow-up studies in postoperative patients indicated that 201Tl brain SPECT could be a useful means of detecting tumor regrowth soon after operation, because CT, MRI and angiography have limited ability to differ-entiate recurrence from necrosis though they can detect ma,ss lesions. 
Whether or not 201Tl SPECT is capable of detect-ing smaller lesions is unclear. However, our exper-ience indicated that it can detect greater than 2 cm in diameter, even if deep seated. L/N ratios and baseline L/N ratio of 2.5 would change to some degree according to the SPECT equipment, filtered back projection algorithm, and precorrection method. Routine use of attenuation correction, although not used in our series, may improve the quantitative assessment of brain tumors. 
In our series, one patient whose 201Tl SPECT changed from negative to positive had tumor recur-rence in an area pathologically proven to be necrotic. In such a case we suspect that the recurrence may have originated in a few remaining viable tumor cells that studded the necrotic tissue. 
Quantitative 201Tl brain SPECT was not always capable of differentiating glioma from other intra-cranial tumors, although low grade astrocytoma and non-malignant lesions including cerebrovascular disease showed little or no tracer uptake. This was in contrast to high 201Tl uptake in glioblastoma. 201Tl brain SPECT cannot therefore be used effec-tively to distinguish glioma from other brain tumors in preoperative patients. Lesions with high 201T1 uptake may refiect glioblastoma or hypervascular tumors, such as meningioma, hemangioblastoma and metastasis from thyroid cancer. But lesion with no 201Tl uptake may reflect low-grade glioma, or non-malignant lesions including cerebrovascular disease. AVM and encephalitis. 201Tl SPECT may be helpful in the histological differentiation of gliomas because glioblastoma is highly suspected when there is visible 201Tl uptake, although there were no preoperative patients with anaplastic astro-cytoma or oligodendroglioma in our series. 
Mountz et al.13 showed from their microauto-radiographic experiment that the mechanism of 201Tl sequestration by high-grade g]ioma is due to its preferential uptake into tumor cells. Preferential tracer uptake involved in the mechanism may reflect an increase in the sodium potassium adenosine triphosphatase (Na+/K+_ATPase) a.ction on viable tumor cell membrane.17 This could in part explain the absence of abnormal 201T1 accumulation in high-grade glioma treated with radiation and/or chemotherapy due to decreased or destroyed cell membrane active transport. On the other hand, both the extent of normal or neovascular blood flow and the integrity of the blood brain barrier appear to be important factors in 201Tl uptake since the majority of tumors with avid tracer uptake on the early images were hypervascular as assessed by CT/MRI or angiography. Minimal uptake of 201Tl by cerebral necrosis is probably related to destruction of the blood brain barrier ; necrotic tissue has neither vascularity nor an intact mechanism of the Na+/K+_ ATPase pump. Conversion from a high to a low 201Tl index between early and delayed images may be dependent in part on the 201Tl washout rate which is also a function of regional blood flow in the lesion; this is the basis for redistribution in 201T1 myocardial scintigraphy. 
The present reassessment of quantitative 201Tl brain SPECT provides support for its use in differen-tiating cerebral necrosis from recurrent tumor estimating residual tumor burden soon after tumor-ectomy, and detecting tumor regrowth earlier in the postoperative patient, since findings in both CT and MRI are usually nonspecific. In preoperative patients, however, quantitative 201Tl brain SPECT cannot be used to distinguish glioma from other brain tumors. 

REFERENCES 

1 . Graeb DA, Steinbok P, Robertson WD : Transient early computed tomographic changes mimicking tumor progression after brain tumor irradiation. Radiology 144: 813-817, 1982 
2. Brismar J, Robertson GH, Davis KR, et al : Radiation necrosis of the brain. Neuroradiological consider-ations with computed tomography. Neuroradiology 12: 109-113, 1976 
3. Martins AN, Johnson JS, Henry JM, et al : Delayed radiation necrosis of the brain. J Neurosurg 47 : 336-345, 1977 
4. Doom GC. Hecht S, Brant-Zawadzki M, et al : Brain radiation necrosis: MR imaging. Radiology 158: 149-155, 1986 
5. Tsuruda JS. Kortman KE, Bradley WG, et al : Radiation effects on cerebral white matter : MR evaluation. AJR 149: 165-171, 1987 
6. Chamberlain MC, Murovic JA, Levin VA : Absence of contrast enhancement on CT brain scans of patients with supratentorial malignant gliomas. Neurology 38: 1371-1374, 1988 
7. Di Chiro G, Oldfield E, Wright DC, et al : Cerebral necrosis after radiotherapy and/or intraarterial chemotherapy for brain tumors : PET and neuro-pathological studies. AJNR 8: 1083-1091, 1987 
8. Doyle WK, Budinger TF. Valk PE, et al : Differen-
262 Shigeru Kosuda, Hirofumi Fujii, Shigeki Aoki, et al 
Annals of Nuclear Medicine tiation of cerebral radiation necrosis from tumor recurrence by F-18-FDG and Rb-82 positron emis-sion tomography. J Comput Assis Tomogr 1 1 : 563-570, 1987 
9. Mineura K, Sasajima T, Kowada M, et al : Innovative approach in the diagnosis of gliomatosis cerebri using carbon-11-L-methionine positron emission tomography. J Nucl Med 32: 726-728, 1991 
10. Ancri DA, Basset JY, Lonchampt MF, et al : Diag-nosis of cerebral lesions by thallium 201. Radiology' 128: 417~,22, 1978 
11 . Kaplan WD, Takvorian T, Morris JH, et al : Thal-lium-201 brain tumor imaging : a comparative study with pathological correlation. J Nucl Med 28 : 47-52, 1987 
12. Black KL. Hawkins RA, Kim KT, et al: Use of thallium-201 SPEC.T to quantitate malignancy grade of gliomas. J Neurosurg 71 : 342-346, 1989 
13. Mountz JM, Raymond PA, McKeever PE, et al : Specific localizatlon of thallium-201 in human high-grade astrocytoma by microautoradiography. Cancer Res 49: 4053-4056. 1989 
14. Kim KT, Black KL. Marciano D, et al : Thallium-201 SPECT imaging of brain tumors: Methods and results. J Nucl Med 3 1 : 965-969, 1990 
15. Schwartz RB. Carvalho PA, Alexander 111 E, et al : Radiation necrosis vs high-grade recurrent glioma : Differentiation by using dual-isotope SPECT with 201Tl and 99mTc-HMPAO. AJR 158: 399-404, 1992 
16. McKusick KA, Fischman AJ, Hochberg FH, et al : Tl-201 and F-18 deoxyglucose in detection of recur-rent primary brain tumor. J Nucl Med 32: p955 (abstract), 1991 
17. Kosarov LB, Friedman H : Enhanced Na+_K+_ activated adenosine triphosphatase activity in trans-forming fibroblast. Cancer Res 34: 1862-1865, 1974