Prostate-Specific Membrane Antigen as Target for Neuroimaging of Central Nervous System Tumors

Introduction Positron emission tomography (PET) imaging with prostate-specific membrane antigen- (PSMA-) binding tracers has been found incidentally to demonstrate uptake in CNS tumors. Following the encouraging findings of several such case reports, there is a growing interest in the potential application of PSMA-targeted PET imaging for diagnostics, theranostics, and monitoring of CNS tumors. This is a systematic literature review on PSMA-binding tracers in CNS tumors. Methods A PubMed search was conducted, including preclinical and clinical reports. One hundred and twelve records were identified, and after screening, 56 were included in the final report. Results Tissue studies demonstrated PSMA expression in tumor vascular endothelial cells, without expression in normal brain tissue, though the extent and intensity of staining varied by anti-PSMA antibody and methodology. Most included studies reported on gliomas, which showed strong PSMA ligand uptake and more favorable tumor to background ratios than other PET tracers. There are also case reports demonstrating PSMA ligand uptake in prostate cancer brain metastases, nonprostate cancer brain metastases, and meningiomas. We also review the properties of the various PSMA-binding radiotracers available. Therapeutic and theranostic applications of PSMA-binding tracers have been studied, including labeled alpha- and beta-ray emitting isotopes, as well as PSMA targeting in directing MRI-guided focused ultrasound. Conclusions There is a potential application for PSMA-targeted PET in neuro-oncology as a combination of diagnostic and therapeutic use, as a theranostic modality for managing CNS tumors. Further research is needed regarding the mechanism(s) of PSMA expression in CNS tumors and its differential performance by tumor type.


Introduction
Standard imaging of central nervous system (CNS) tumors with magnetic resonance imaging (MRI) has demonstrated limitations [1], and positron emission tomography (PET) has emerged as an additional prominent noninvasive imaging modality for CNS tumors. Among the PET tracers under investigation for imaging of CNS tumors is prostate-specific membrane antigen (PSMA) binding tracer, whose utility in the CNS was originally noted on PET scans of patients with prostate cancer [2][3][4]. PSMA is a type II transmembrane glycoprotein that was originally identified due to its high expression in malignant prostate vasculature. Subsequent research has revealed that it is also expressed in a variety of other tissue types in different organ systems [2,4,5]. PSMA is also known as glutamate carboxypeptidase II (GCPII) in the brain, as well as folate hydrolase I in the intestines [6], where its main enzymatic function is to cleave N-acetyl-L-aspartyl-L-glutamate (NAAG) to N-acetylaspartate and glutamate [7]. In the normal rodent and human brain, astrocytes demonstrate some GCPII expression, which provides the majority of NAAGhydrolyzing activity [6,8], implicating this enzyme in some neuropsychiatric conditions in which glutamate is considered neurotoxic. As a result, inhibition of GCPII was thought to be neuroprotective, and GCPII inhibitor administration increased brain NAAG levels, improving cognitive performance in mice with experimental autoimmune encephalomyelitis [9]. Interestingly, efforts to develop small molecule inhibitors of GCPII to treat brain disorders of hyperglutamatergic pathogenesis led to the discovery of their application outside of the nervous system, most prominently as imaging and therapeutic agents in prostate cancers [10].
PET radioligands targeting PSMA are misleading in their nomenclature as "PSMA tracers," because they are in reality inhibitors with high affinity for the PSMA-binding motif [11]. Such PSMA-targeting tracers have in recent years garnered greater interest because of their potentially widespread oncologic diagnostic applications, as well as therapeutic applications including those recently developed for prostate cancer. Importantly, PSMA is shown to have specific expression on tumor vasculature, which is unique from other vascular imaging tracers which nonspecifically bind to the vasculature in both normal tissue and tumor tissue, and this distinction may therefore facilitate direct targeting of tumor angiogenesis [12]. Case reports initially revealed the incidental finding of PSMA accumulation in CNS metastases, leading to subsequent exploration of PSMA expression and imaging in a variety of CNS malignancies. Here, we systematically review the literature investigating PSMA-targeted tracers for CNS lesions, including all identified in vitro and in vivo studies, in order to summarize current knowledge regarding the potential for PSMA-targeted PET in neurooncology imaging.
A total of 101 records were returned. The bibliographies of these papers were reviewed to elicit additional papers, yielding 11 additional articles. These 112 reports were then screened for original research publications related to PSMA and neuro-oncology, including preclinical studies, retrospective and prospective clinical studies, and case reports, but excluding review papers, systematic reviews, and metaanalyses. This yielded 56 studies for inclusion in the final review. See the study selection flowchart in Figure 1.
Data collection and extraction were performed independently by one author (BS) with oversight by the other authors. Data variables collected included year, study design, study subjects, tumor type, tracer name, and main study results. Data were analyzed and summarized qualitatively.

Expression of PSMA in Tissue and Preclinical Models.
There are 13 published reports of PSMA expression in CNS tumor tissue and preclinical models, including a collective total of 331 patients and 38 animals ( Table 1). The first such report was by Chang et al. (1999), in which the expression profile of several PSMA antibodies in a wide variety of tumor tissue types, among them a single glioblastoma (GBM), were examined [13]. They reported that GBM neovasculature cells stained PSMA-positive, while neither tumor cells nor normal brain demonstrated expression. Wernicke et al. (2011) published an immunohistochemistry study of 32 GBMs, using an anti-SMA mAb 3E6 (Dako) stain, quantifying the extent and intensity of vascular endothelial staining [14]. They reported that all GBMs stained positive for PSMA, with variable extent and intensity. The extent of staining was mostly 51-100%, while the intensity of staining was mostly moderate or maximum. The PSMA staining colocalized within the areas of tumor with CD31, a known blood vessel marker, while, again, no staining was seen in any normal brain tissue.
The same group then published in 2014 a PSMA immunohistochemistry study of 14 breast cancer patients with brain metastases [15]. They used a mouse 3E6 anti-PSMA antibody (Dako) for immunohistochemistry staining and quantified the extent of staining in tumor-associated vessels. The tumor vasculature of all brain metastases stained PSMA-positive, and the extent of expression was greater than 50%. They also scored the primary tumor for 10 of these patients, and in all cases, the primary tumor had greater than 50% expression. That same year, Nomura et al. published a PSMA tissue study of 23 gliomas and breast cancer brain metastases [16]. They used a mouse PSMA antibody mAb 3E6 (Dako) for immunohistochemistry and quantified the intensity of PSMA staining in tissue relative to a staining calibration curve normalized to mean image background intensity, using the automated algorithms in the Scanscope® CM-1 scanner and ImageScope® software. They found that GBM blood vessels stained heavily, grade II/III gliomas showed some (<2%) tumor tissue staining but no vessel staining, and grade I gliomas showed moderate vessel staining and some tumor staining. They additionally found that normal tissue blood vessels did not stain for PSMA but fewer than 5% of normal neurons did. The quantified relative intensity of staining in all gliomas was statistically significantly higher than in normal brain. They also reported corresponding PSMA and von Willebrand factor 2 Molecular Imaging (VWF) staining on vascular endothelial cells. In brain metastases, they found variable staining within and between tumors, but on average the staining was significantly greater than normal brain. The above data were complemented by several subsequent case reports. Schwenck et al. reported increased PSMA expression in the vascular endothelium of a GBM patient, though not in the normal brain and vasculature ( Figure 2) [17]. Subsequently, Unterrainer et al. reported the first PSMA study in a gliosarcoma patient, which demonstrated strong PSMA expression in the neovascular endothelial cells and not in the tumor tissue cells [18]. Salas Fragomeni et al. reported in one anaplastic astrocytoma and 2 GBMs that PSMA staining was localized to the vascular cells in GBM and tumor cells in anaplastic astrocytoma, and no staining was found in normal brain or vessels [19].
The two largest of these studies, by Matsuda et al. [20] and Saffar et al. [21] were both published in 2018 and demonstrated divergent results regarding the expression of PSMA in high-grade gliomas. Matsuda et al. studied tissue samples from 78 glioma, brain metastasis, CNS lymphoma, and radiation necrosis patients [20]. They used a rabbit monoclonal anti-PSMA antibody (EPR6253, Abcam) for immunohistochemistry, and they quantified the strength of tissue PSMA expression in vascular endothelial cells. They reported that PSMA was expressed in the vascular endothelial cells of almost all GBMs, most grade I and III gliomas, and all metastatic brain tumors. However, PSMA was expressed in few grade II gliomas and primary central nervous system lymphomas (PCNSL). They found no PSMA expression in radiation necrosis tissue. GBMs and brain metastases showed the highest expression levels, while few grade III gliomas and no grade II gliomas showed high expression. Saffar et al. reported on PSMA expression in 72 gliomas [21]. They used a monoclonal liquid Novocastra™ mouse monoclonal antibody (clone 1D6, Novocastra) for immunohistochemistry. They quantified the extent and intensity of PSMA staining in vascular endothelial cells. They found positive PSMA staining in a minority of gliomas, regardless of tumor grade. Among these, the GBMs and grade I gliomas had a lower extent of vascular staining, while the grade II and III gliomas had a higher extent of staining. The intensity of staining was weak for most of the GBMs and the grade III glioma, and it was moderate for the grade I and II gliomas.
The following year, Mahzouni et al. published a larger PSMA tissue study in GBMs (n = 60) [22]. They used an anti-PSMA mAB clone SP29 (Biogenex) for immunohistochemistry and quantified the extent and intensity of PSMA staining in vascular endothelial cells. They found PSMA     68 Ga-PSMA and 18 F-DCFPyL binding were studied on ex vivo autoradiography in 38 rats, implanted with either F98, 9L, or U87 glioma cells [23]. They found that both PSMA-targeting tracers exhibited strong binding in the peritumoral area but moderate binding in the tumor core. In vivo animal PET imaging showed a higher tumor to background ratio (TBR) for 18 F-DCFPyL (TBR 6.28-7.92) than 68 Ga-PSMA (TBR 3.22-3.92). Tissue staining with three different anti-PSMA antibodies showed heterogeneous results, with one antibody negative for all three tumor cell lines (ab58779, Abcam), one positive in all three (NBP1-45057, Novus), and one split (NBP1-89822, Novus). PSMAspecific binding was confirmed by application of the PSMA-antagonist PMPA which effectively suppressed 68 Ga-PSMA and 18 F-DCFPyL binding. They found that activated microglia expression (CD11b) was low intratumorally and peritumorally but activated astrocyte expression (GFAP) was high peritumorally. They concluded that PSMA expression may truly be capturing astrocyte activation instead of tumorigenesis, which would limit its utility in differentiating tumor recurrence from radiation necrosis.
Recently, Liu et al. published a report from 30 glioma patients, in which they demonstrated PSMA expression on IHC staining in zero out of 14 grade II gliomas, two out of four grade III gliomas, and nine out of 12 GBMs [24]. Then, Holzgreve et al. demonstrated, in a series of 16 GBMs, using mouse mAb 3E6 (Agilent), that all 16 had PSMA uptake on IHC at initial diagnosis and 15/16 at recurrence [25]. The change in PSMA expression varied between these timepoints, whereas the vessel marker CD34 remained consistent. The level of vascular PSMA expression at recurrence was predictive of survival, as was an increase in PSMA expression over the course of the disease. Notably, PSMA expression was not associated with MGMT status or Ki-67 proliferation index.
In summary, tissue from multiple types of CNS tumors demonstrates PSMA expression. GBM and breast cancer brain metastases tissue showed robust PSMA staining, localized to the neovasculature, with staining present in a majority of cells with moderate to maximum intensity. PSMA staining metrics in GBM were also predictive of survival. Glioma tissue PSMA staining was highly variable across studies. No PSMA staining was seen in radiation necrosis human tissue; however, in glioma cell lines, there was evidence that PSMA expression may capture astrocyte activation instead of tumorigenesis.

PSMA-Targeting PET Tracer
Properties. The properties of PSMA-targeting tracers differ on several key features, which results in different strategies for targeting PSMA [26]. For example, some tracers are small-molecule inhibitors while others are monoclonal antibodies [27]. The small-molecule tracers are smaller than antibody tracers and thus can travel faster throughout the vasculature and are excreted faster. Therefore, small-molecule tracers reach more tissue indiscriminately but allow for faster imaging protocols. Antibody tracers travel preferably through larger vessels, which are often seen more in tumor tissue than normal tissue. Therefore, antibody tracers allow for more specific imaging although the scanning protocol is longer. Both of these types of targeting compound are conjugated with a radionuclide to create a radiotracer targeting PSMA.
There is variability between radionuclides, including their half-life and positron range. These tracer characteristics are described in detail below.

Gallium-Tagged Tracers.
Four formulations of 68 gallium-(Ga-) tagged PSMA-targeting tracer have been developed, including 68 Ga-PSMA-11 (HBED-CC), 68 Ga-PSMA-617, 68 Ga-PSMA-I&T, 28 and 68 Ga-THP-PSMA (Table 2) [28]. The 68 Ga radionuclide is produced in a Ga generator, has a half-life of 1.1 hours, a positron range of 8.9 mm, and is taken up by the bladder wall and kidney [29,30]. They do not show uptake in normal brain tissue. They may have a lower resolution than other isotopes, due to a longer photon range and the energy within the isotope. 68 Ga-PSMA-11 (HBED-CC) benefits from a high affinity for PSMA-expressing tumors. It also shows rapid blood clearance and lower liver uptake, though it has uptake in salivary glands [29]. PSMA-617 can be labeled with 177 Lu, 255 Ac, or 90 Y, for theranostic pairing, which gives it the potential to extend beyond the diagnostic sphere and into the treatment of PSMA-expressing lesions. 68 Ga-PSMA-Imaging & Therapy (I&T) has high-affinity tumor uptake   (Table 2). Among these tracers, only 68 Ga-PSMA-11 (HBED-CC) and 68 Ga-THP-PSMA have been studied in CNS tumors to date.

Fluorine-Tagged
Tracers. Three formulations of 18 fluorine-(F-) tagged PSMA-targeting tracers have been developed, including 18 F-DCFPyl [31], 18 F-PSMA-1007 [31], and 18 F-DCFBC [32]. These 18 F radionuclides are produced in a cyclotron, have a half-life of 1.8 hours, a positron range of 0.6 mm, and are not taken up by normal brain tissue [30][31][32]. The route of excretion and critical uptake organs vary by tracer, and they include the liver, kidney, bladder wall, and gallbladder. These tracers all benefit from a better resolution than other isotopes, due to their short photon range. However, they all have significant salivary gland uptake. 18 F-DCFPyL has a higher target-to-background ratio than 68 Ga-PSMA-11 (HBED-CC) and thus may detect more lesions [29]. It has a much higher binding affinity for PSMA than its predecessor, 18 F-DCFBC, which also suffered from high blood pool activity which limited detection of lymph nodes near blood vessels. 18 F-PSMA-1007 is structurally  68 Ga-PSMA-617 but potentially has a better resolution [31]. When there are low PSA levels, subcentimeter bone lesions, lymph node involvement, or hepatic involvement, 68 Ga-PSMA agents are superior to 18 F-choline (Table 2). Among these, only 18 F-DCFPyl and 18 F-PSMA-1007 have been studied in CNS tumors to date.

3.2.3.
Copper-Tagged Tracers. One 64 copper-(Cu-) tagged PSMA-targeting tracer was developed, and it was based on the structural base of 68 Ga-PSMA-617 [33]. The 64 Cu radionuclide is produced in a cyclotron, has a long half-life of 12.7 hours, a positron range of 0.6 mm, and accumulates in the liver, large intestine, and pancreas [30,33]. It is not taken up by normal brain tissue. It benefits from theranostic pairing potential with 67 Cu (Table 2). This radiotracer has not yet been studied in CNS tumors.

Iodine-Tagged
Tracers. There has been one 124 iodine-(I-) tagged PSMA-targeting tracer developed, 124 I-MIP-1095 [34]. The 124 I radionuclide is produced in a cyclotron, has a long half-life of 100.8 hours, a positron range of 3.4 mm, and accumulates in the salivary glands, liver, and kidneys [30,34]. It is not taken up by normal brain tissue. It benefits from theranostic pairing potential with 131 I ( Table 2). This radiotracer has not yet been studied in CNS tumors.

Zirconium-Tagged Tracers.
There has been one 89 zirconium-(Zr-) tagged PSMA-targeting tracer developed, 89 Zr-Df-IAB2M [35]. The 89 Zr radionuclide is produced in a cyclotron, has a long half-life of 78.4 hours, a positron range of 1.2 mm, and accumulates in the liver and kidney [35,36]. It is not taken up by normal brain tissue. This radiotracer has been studied in CNS tumors, but further clinical investigations are needed to understand its advantages and disadvantages relative to other tracers ( Table 2).
3.2.6. Technetium-Tagged Tracers. There have been five 99m technetium-(Tc-) tagged PSMA-targeting tracers developed, including Tc-PSMA [37], Tc-MIP-1404 [38], Tc-MIP-1405 [39], Tc-PSMA-I&S [40], and Tc-EDDA/ HYNIC-iPSMA [41]. These radionuclides are produced by 99m Tc-generator and have a half-life of 6 hours [37][38][39][40][41]. The route of excretion and uptake by critical organs varies by tracer, but the kidneys are involved in all five, and none show uptake in normal brain tissue. All five Tc-tagged PSMA-targeting tracers have limited clinical application. As single-photon emission computerized tomography (SPECT) imaging tracers, they have a lower resolution than PET imaging tracers, although SPECT is more readily available in clinical settings than PET. Tc-PSMA-Imaging & Surgery (I&S) is being investigated for potential use in targeted surgery ( Table 2). These radiotracers have not been studied in CNS tumors and are unlikely to be developed for this purpose given the superiority of PET over SPECT with regard to resolution.  Table 3). Most of these studies have focused on imaging with the 68 Ga-PSMA radiotracer, though a few have used alternative tracers such as 18 F-DCFPyL, 89 Zr-Df-IAB2M, and 18 F-PSMA-1007. Following the promising results of early immunohistochemistry studies in glioma tissue, the first report of PSMA-targeting PET imaging of a glioma patient was published by Schwenck et al. [17]. They found that 68 Ga-PSMA-11 (HBED-CC) uptake was markedly increased in the contrast-enhancing solid tumor region identified on MRI and that there was no uptake in the normal brain tissue (Figure 2). Then, Unterrainer et al. published the first case report of 68 Ga-PSMA PET imaging in a gliosarcoma patient [18]. They found high 68 Ga-PSMA-11 (HBED-CC) uptake, with a median maximal standardized uptake value (SUV max ) of 3.43 and median maximal TBR (TBR max ) of 48.93.

Diagnostic Performance in
The first case series of 68 Ga-PSMA PET imaging in GBM was published by Sasikumar et al., which included five GBMs with suspected recurrence and one newly diagnosed GBM [42]. In four out of five GBMs with suspected recurrence, they found increased uptake of both 68 Ga-PSMA-11 (HBED-CC) and 18 F-FDG correlating to the lesion on MRI with histological confirmation. While the uptake of both tracers was spatially correlated, TBR was greater for 68 Ga-PSMA-11 (HBED-CC) (12.9) than for 18 F-FDG (0.96). The 5 th suspected recurrence, which did not demonstrate uptake with either tracer, did not have recurrence on histology. The newly diagnosed GBM showed intense tracer uptake on the periphery of the lesion, with a TBR of 22.3 for 68 Ga-PSMA-11 (HBED-CC) and 1.11 for 18 F-FDG. In this case series, they noted a better visualization of the lesion using 68 Ga-PSMA-11 (HBED-CC) than 18 F-FDG, which they attribute to its comparatively greater TBR. Sasikumar et al. later published a larger case series of 15 gliomas [43], which included the six GBMs from their 2017 report. In their 2018 report, they found that nine out of 10 suspected recurrences had positive findings on 68 Ga-PSMA-11 (HBED-CC) PET scan, and the one without 68 Ga-PSMA-11 (HBED-CC) uptake had no evidence of recurrence on further testing. The TBR values ranged from 4.07 to 29.4, compared to the patient with no disease whose TBR value was 1.15. They additionally found increased 68 Ga-PSMA-11 (HBED-CC) uptake in two newly diagnosed GBMs and two postsurgical GBMs, but no uptake in a postsurgical grade III oligodendroglioma. Overall, in 13 68 Ga-PSMA-11 (HBED-CC) positive gliomas, the TBR was 34.78 in grade II, 11.9 and 27.0 in grade III, and 4.07-134.8 in grade IV. They concluded that the tracer uptake did not correlate to glioma grade ( Figure 3).
A case report from Kunikowska et al. demonstrated high 68 Ga-PSMA uptake in a GBM patient, with a SUV max of 23.7 [44]. A report of 68 Ga-PSMA PET imaging in an oligodendroglioma from Malik et al. [45] demonstrated increased 68 Ga-PSMA uptake in the lesion and better lesion delineation with 68 Ga-PSMA than with 18 F-FDG. Verma et al. published a case series of 10 glioma patients, all of which demonstrated increased uptake on 68 Ga-PSMA-11 (HBED-CC) PET imaging [46]. They found the SUV max among 9 Molecular Imaging  Ga-PSMA (i) Moderate correlation between tumor grade and SUV max (r = 0:53), SUV mean (r = 0:55), SUV peak (r = 0:50) (ii) Grade II/III gliomas had significantly lower SUV max than GBMs, with a cutoff of 2.3 (iii) LGG versus HGG cutoff for SUV max was 1.15 (iv) 68 Ga-PSMA was more sensitive (p < 0:0%) than MRI, but not more specific GBM patients (16:93 ± 5:4) to be significantly higher than that of grade II gliomas (2:93 ± 0:3), as well as the TBR  68 Ga-PSMA-11 (HBED-CC) uptake on PET imaging of a posttreatment GBM with suspected recurrence, which was ultimately diagnosed as pseudoprogression based on subsequent imaging. The lesion SUV max was 2.71, versus 0.52 in normal brain, and a TBR of 5.21. They warn that this false positive indicates that 68 Ga-PSMA may not be fully able to differentiate recurrence from radiation necrosis. Moreau et al. echoed these concerns in their case report of a GBM patient with modest 68 Ga-PSMA-11 (HBED-CC) uptake (SUV max of 3.2), who was determined to have postradiation changes based on clinical imaging [48]. Without an established threshold for defining recurrence versus postradiation treatment effect, it may be difficult to interpret some of the 68 Ga-PSMA PET images, especially given the wide range of values for SUV max and TBR that have thus far been demonstrated in gliomas.
Pernthaler et al. report an oligodendroglioma patient with homogenously high 68 Ga-PSMA-11 (HBED-CC) uptake and 18 F-fluciclovine uptake, on PET imaging [49]. The SUV max of 68 Ga-PSMA-11 (HBED-CC) (9.7) was greater than that of 18 68 Ga-PSMA uptake which was low in the core of the lesion and higher within foci along the edge of the lesion. On histopathology, it was found that the low uptake areas were grade II glioma tissue while the high uptake areas were grade III glioma tissue [52].
Kunikowska et al. published the largest series of GBMs imaged with 68 Ga-PSMA-11 (HBED-CC) PET (n = 15) [53]. They found 100% spatial correlation of 68 Ga-PSMA-11 (HBED-CC) uptake with MRI lesions, and 100% of the patients demonstrated increased 68 Ga-PSMA-11 (HBED-CC) uptake, though the intensities and patterns differed. The median SUV max was 6.5 (range 2.1-14.3), SUV mean was 3.5 (range 1.3-6.1), and TBR was 96.7 (range 32.2-357.5). This TBR range is higher even than the range reported by Sasikumar et al. [43] which indicates a high degree of differentiation between tumor and normal brain. These values are also higher than the reported TBRs of amino acid PET tracers like 18 F-FET and 11 C-MET PET [54], which demonstrates that 68 Ga-PSMA better differentiates lesions from the background. Kunikowska et al. also investigated the tumor-to-liver (TLR) ratio because a TLR of 1.5 or higher is needed to be eligible for current PSMAbased radionuclide targeted therapy. They found the median TLR was 0.8 (range 0.6-1.8), and only 2/15 (13%) patients had a TLR of 1.5 or higher.
In a study of 30 pathology-confirmed glioma patients, Liu et al. found that SUV max and SUV mean were higher for 68 Ga-PSMA PET (AUC 0.96 and 0.94) than for FDG PET (AUC 0.79, 0.74) [24]. 68 Ga-PSMA PET was more effective than FDG PET for distinguishing HGG from LGG. 68 Ga-PSMA SUV max for LGG was 0.74 and for HGG was 5.8, while FDG SUV max was 9.2 for LGG and 11.8 for HGG. With a cutoff of 2.21, 68 Ga-PSMA SUV max achieved sensitivity 0.81, specificity 1.00, and p < 0:001, while with a cutoff of  HGGs, which all showed increased 18 F-DCFPyL uptake [19]. The SUV max ranged from 5.8 to 13.5 in tumor lesions, and no uptake was seen in normal brain tissue. Matsuda et al. used 89 Zr-Df-IAB2M PET to image 2 HGGs [20]. One HGG showed high uptake in the areas of robust contrast enhancement and low uptake in the areas of lesser enhancement. The other HGG showed high but heterogeneous uptake within the lesion, but with a different distribution pattern than 11 C-MET uptake. Marafi et al. evaluated 1 recurrent GBM patient with 18 F-PSMA-1007 PET [56]. This patient showed increased uptake of both 18 F-PSMA-1007 and 18 F-FDG in the area of the lesion on MRI, though the differential uptake of 18 F-PSMA-1007 in tumor versus normal brain was better than 18 F-FDG, so the lesion was more clearly delineated with 18 F-PSMA-1007.

PSMA in Gliomas Summary.
In summary, 68 Ga-tagged PSMA-targeting PET showed robust uptake in gliomas and GBMs, with no uptake in normal brain regions. Across glioma grades, these tracers showed greater uptake than 18 (Table 4). In nearly all cases, these were detected incidentally on 68 Ga-PSMA screening for metastases. In some reports, the diagnosis of prostate cancer metastasis was made using MRI. The first was published in 2015 by Chakraborty et al., in which they reported a prostatic adenocarcinoma patient without neurologic symptoms, who demonstrated avid 68 Ga-PSMA-11 (HBED-CC) uptake in three separate MRI-diagnosed brain metastases, in the cerebellar hemisphere and bifrontal lobes [57]. Dureja et al. reported a prostate adenocarcinoma patient without neurologic symptoms who was found on 68 Ga-PSMA-11 (HBED-CC) PET imaging to have at least four MRI-diagnosed brain metastases in the frontal and frontoparietal lobes [58]. The SUV max of the largest lesion was 4.0. Langsteger et al. reported the only MRI-diagnosed spinal cord metastasis from prostate cancer detected with 68 Ga-PSMA PET imag-ing [59]. The patient had no neurologic symptoms but was found to have 68 Ga-PSMA uptake in the spinal cord at the T4-T10 levels.
In other reports, the diagnosis was confirmed by histopathology. Chan et al. reported a patient with prostate cancer who presented with mild ataxia and nausea and upon imaging with 68 Ga-PSMA-11 (HBED-CC) was found to have a single metastasis in the cerebellum [60]. The 68 Ga-PSMA-11 (HBED-CC) uptake in the lesion was SUV max 5.9, and diagnosis was confirmed by histopathology consistent with high-grade metastatic prostate adenocarcinoma with positive staining for PSMA. Ross et al. reported an asymptomatic temporal lobe metastasis which was not detected on conventional imaging, but was found on 68 Ga-THP-PSMA PET imaging and was confirmed with histopathology to be prostate adenocarcinoma [61]. Yin et al. (2019) reported a patient with prostate adenocarcinoma without neurologic symptoms who was found incidentally to have temporal and occipital metastatic lesions on 68 Ga-PSMA PET [62]. The lesions demonstrated 68 Ga-PSMA avidity with SUV max ranging 3-11, and not all of the metastases were detected on MRI. Histopathology confirmed the diagnosis and demonstrated positive staining for PSA. Of note, this is reported to be the longest surviving prostate cancer patient with multiple brain metastases at three years, which the authors attribute to the early detection with 68 Ga-PSMA PET.
In two reports, the application of PSMA therapeutics reduced both the size of the metastases and the 68 Ga-PSMA uptake on PET imaging. Wei et al. reported two prostate cancer patients with known brain metastases who underwent 68 Ga-PSMA-11 (HBED-CC) PET imaging for restaging and demonstrated distinct 68 Ga-PSMA-11 (HBED-CC) uptake in the cerebral lesions [63]. The 68 Ga-PSMA-11 (HBED-CC) uptake in these two patients decreased as the size of the lesions regressed, following combined 177 Lu-PSMA-617 with radiotherapy. Sathekge et al. reported one castration-resistant prostate cancer patient with several cerebral metastases that showed avid 68 Ga-PSMA uptake on PET imaging [64]. Following treatment with 225 Ac-PSMA-617, the size and 68 Ga-PSMA uptake of the cerebral metastases decreased, along with the PSA levels. On the other hand, Parihar et al. reported a case of new brain metastases with 68 Ga-PSMA uptake, after treatment of prostate cancer with 177 Lu-PSMA and 225 Ac-PSMA [65]. This patient was treated with docetaxel, cabazitaxel, enzalutamide, and two cycles of 177 Lu-PSMA therapy, and then four cycles of 225 Ac-PSMA therapy. After two cycles of 225 Ac-PSMA therapy, 68 Ga-PSMA PET/CT imaging did not demonstrate any brain lesion, and then after four total cycles of 225 Ac-PSMA, imaging revealed new brain lesions in the setting of generalized disease progression.
The first extended case series of PSMA-targeting PET in prostate cancer brain metastasis (n =8), by McBean et al., revealed considerable variability across patients [66]. Only three out of eight patients presented with neurological symptoms, and the 68 Ga-PSMA avidity ranged from not avid to intense, while the SUV max ranged from 2 to 21. Of note, four out of eight patients had previously undergone 177 Lu-PSMA therapy for their primary prostate cancer, but 13 Molecular Imaging this did not correlate with any trends in their imaging findings.

Brain Metastases from Nonprostate
Cancer. There has also been a selection of case reports of brain metastases that arise from nonprostate cancers that nevertheless demon-strate significant PSMA-targeted tracer uptake on PET (Table 5). Four such reports of PSMA-ligand uptake in brain metastases from breast cancer patients have been published. Medina-Ornelas et al. reported a patient with HER-2positive breast carcinoma, who presented with progressive headache and dizziness, and was subsequently found to have Table 4: Summary of PSMA-targeted imaging in prostate cancer brain metastases.

Authors
Year Ga-PSMA (i) 68 Ga-PSMA uptake in new brain metastases, after 177 Lu-PSMA and 225 Ac-PSMA therapy  RCC patient with three CT-diagnosed brain metastases, in the parietal and temporal lobes, with 18 F-DCFPyL SUV max ranging from 0.5 to 3.4 [71].
There is only one report of 68 Ga-PSMA uptake in a brain metastasis from melanoma. Hod et al. reported a patient who demonstrated a melanoma metastasis in the occipital lobe on CT and MRI and unexpectedly showed avid 68 Ga-PSMA uptake in the lesion [72]. Diagnosis was confirmed by histopathology.
Similarly, there has been only one report of 89 Zr-Df-IAB2M uptake in brain metastases from lung cancer. Matsuda et al. report a lung cancer patient with MRI-detected occipital brain metastasis, which demonstrated avid 89 Zr-Df-IAB2M uptake [20]. There was no uptake in normal tissue. Diagnosis was confirmed by histopathology, and immunohistochemical analysis demonstrated a trend toward a positive correlation between 89 Zr-Df-IAB2M uptake and PSMA expression.

Meningiomas.
PSMA targeting is potentially well suited to meningiomas, given that these lesions are generally highly vascularized. There have been eight single case reports published of PSMA-targeted PET imaging in meningioma patients, all of which were incidental findings in patients with prostate cancer (Table 6). Often the lesions were suspected brain metastases that were ultimately diagnosed as meningiomas. In six of these reports, the PSMA-binding tracer utilized was 68 68 Ga-PSMA uptake, and diagnosis was supported by MRI [75]. Courtney et al. reported a patient with a history of prostate cancer who demonstrated 68 Ga-PSMA uptake in the left frontal lobe, confirmed as a preexisting meningioma [76]. Finally, Junqueira et al. reported avid 68 Ga-PSMA uptake (SUV max 12.1) in an MRIdetected lesion, suggestive of an intraventricular meningioma of a patient with history of prostate cancer [77].
One report utilized the 64 Cu-PSMA tracer. Calabria et al. reported an MRI-detected foramen magnum meningioma with SUV max of 3.8 at 1 hour after tracer administration and 3.9 at four hours postadministration [78].
Similarly, one report utilized the 18 F-PSMA-1007 tracer. Haemels et al. reported a patient with moderately avid 18 F-PSMA-1007 uptake in the occipital lobe [79]. This was diagnosed by histopathology as a transitional-type meningioma. Tissue autoradiography studies revealed specific 18 F-PSMA-1007 binding, which could be inhibited by 2-PMPA.
3.4.5. PSMA in Other CNS Tumors Summary. In summary, PSMA-targeting PET showed uptake in prostate cancer CNS metastases with variable avidity. Only case reports and small case series have been published though, and there is a need for systematic study of this imaging modality in this patient population. Anecdotal PSMA-targeted PET imaging has also shown uptake in brain metastases from breast cancer, RCC brain, melanoma, and lung cancer. There are also a few reports of 68 Ga-tagged PSMA-targeted tracer uptake in meningiomas, though more research is needed to understand the potential use of this imaging modality in meningiomas.

Potential Therapeutic and Theragnostic
Applications of PSMA 3.5.1. Labeled Alpha-Ray Emitting Radioisotopes. 225 Ac-PSMA-617 is an alpha-emitting radioisotope-labeled derivative of PSMA-617, which targets the overexpressed PSMA seen in prostate cancer, and this therapy has demonstrated excellent therapeutic response in prostate cancer [82,83]. It was also recently noted to be effective in the treatment of prostate cancer brain metastases [64]. Initial assessment with 68 Ga-PSMA PET revealed cerebral metastases, and then, following one and two cycles of 8 MBq of 225 Ac-PSMA-617 treatment, there was a functional response on 68 Ga-PSMA PET. The serum PSA level dropped from an ini-tial 788.63 μg/L to 6.52 μg/L, to 0.32 μg/L. This incidental finding demonstrates the potential for this prostate cancer therapy to be applied in brain metastases and perhaps other PSMA-expressing CNS tumors as well ( Figure 6).

Labeled
Beta-Ray Emitting Radioisotopes. 177 Lu is a beta-emitting radioisotope, and 177 Lu-PSMA-617 targets overexpressed PSMA and has demonstrated excellent therapeutic response in metastatic prostate cancer [84]. The first report of 177 Lu-PSMA-617 therapy use in prostate cancer brain metastases came from Wei et al., in which they report significant regression of cerebral lesions following 177 Lu-PSMA-617 treatment plus local radiotherapy in two patients [63]. The first had several radiation-refractory cerebral and cerebellar metastases, which showed 68 Ga-PSMA-11 (HBED-CC) uptake on initial PET imaging. After four cycles of 177 Lu-PSMA-617 therapy totaling 25.5 GBq, the lesions demonstrated reduced size and PSMA expression. This correlated to a decrease in serum PSA from 195 ng/mL to 2.4 ng/mL. The second patient had one brain metastasis that was diagnosed after several hormonal and chemotherapy systemic treatments. This lesion showed 68 Ga-PSMA-11 (HBED-CC) uptake on initial PET, and after three cycles of 177 Lu-PSMA-617 therapy, the first of which was combined with radiotherapy; it had nearly completely regressed. This was accompanied by a decrease in serum PSA from 112 ng/mL to 3.9 ng/mL. Recently, Kumar et al. reported on the successful application of 177 Lu-PSMA-617 therapy in a glioblastoma patient [51]. The patient presented with recurrent disease, which demonstrated intense 68 Ga-PSMA-11 (HBED-CC) uptake in the MRI-enhancing areas. The patient underwent three cycles of 177 Lu-PSMA-617 for a total dose of 3700 MBq, after which MRI and 68 Ga-PSMA-11 (HBED-CC) PET demonstrated a reduction in lesion size (from 18 mL to 5.4 mL), which was accompanied by an improvement in ECOG score from 4 to 3. These positive results are tempered by a recent report from Parihar et al., which showed the emergence of new brain Ac-PSMA treatment of prostate cancer [65].

MRI-Guided Focused
Ultrasound. MRI-guided focused ultrasound (MRgFUS) has been leveraged as a tool for disrupting the blood-brain barrier (BBB) in order to permit the passage of drugs, for a number of CNS indications [85]. In a preclinical study of nine rats with nondiseased brains, Airan et al. demonstrated that 18 F-DCFPyL could be administered and targeted to a specific brain focus using MRgFUS for BBB disruption [86]. Following MRgFUS, 18 F-DCFPyL uptake localized to the foci of BBB disruption and extended minimally beyond the contrast-enhancing MRI lesion. Target uptake plateaued at 80 minutes postadministration, with a maximal TBR of 7. PSMA-binding specificity was confirmed by administering the anti-PSMA ZJ-43 which lowered TBR fourfold. Although this was performed on healthy brains, it demonstrated the possibility of leveraging MRgFUS for opening the BBB to allow hydrophilic 18 F-DCFPyL PSMA to enter and reach target tissue within the brain.

Future Objectives
The PSMA-binding tracer mechanism of expression is thought to be specific to tumor vasculature, not to invading or proliferating cells. While this may limit its utility as an isolated imaging modality, it may prove to be a useful adjunct to other neuroimaging modalities for CNS tumors.
On the other hand, the significant potential advantage of PSMA-targeted PET tracers is in CNS tumor theranostics, which the current literature only begins to describe. There may be the potential for a wide range of future applications, such as small molecule selective PSMA inhibitors and radiosensitizers. Pairing PSMA-targeted diagnostic imaging with PSMA-targeted therapeutics may yield a robust theragnostic option for CNS tumor patients.

Conclusions
In summary, this systematic review of the literature on PSMA-targeted tracer use in CNS tumors demonstrates promising results about the potential for this PET tracer for neuro-oncologic imaging. Future efforts should continue to explore the potential for PSMA in neurooncologic imaging, specifically focusing on proposed mechanism(s) of PSMA expression in CNS tumors, differential imaging performance by CNS tumor type, direct comparison of PSMA-targeted PET to other neuroimaging modalities for diagnosing and monitoring CNS tumors, and direct comparison of competing PSMA-binding radiotracers. The surreptitious discovery of the utility of PSMAtargeted PET imaging in CNS tumors has invigorated much excitement in the field about its potential clinical application, but more research is needed to better understand this PET tracer before it can be implemented clinically.

Data Availability
The data supporting this systematic review article are from previously reported studies included within the article and which have been cited.