Somatostatin Receptor Subtype 2 Is Expressed by Supratentorial Primitive Neuroectodermal Tumors of Childhood and Can Be Targeted for Somatostatin Receptor Imaging

Vol. 10, 2997–3006, May 1, 2004 Clinical Cancer Research 2997

Michael C. Fru¨hwald,1 Christian H. Rickert,2,3 (stPNET), 1 medulloepithelioma (ME), and 8 ependymomas M. Sue O’Dorisio,4 Mark Madsen,5 were screened. Tumors positive in vitro were additionally Monika Warmuth-Metz,6 Geetika Khanna,5 analyzed in vivo using SRI. 2 7 Results: Abundant expression of somatostatin receptor Werner Paulus, Joachim Ku¨hl, type 2 in stPNET, a ME, and ependymomas warranted in 1 8 Heribert Ju¨rgens, Peter Schneider, and vivo imaging of 7 stPNET, 1 rhabdomyosarcoma, 3 ependy- Hermann L. Mu¨ller9 momas, 1 ME, and 1 glioblastoma. Although SRI was pos- 1Department of Pediatric Hematology and Oncology, University itive in 6/7 stPNET, 1 rhabdomyosarcoma, and 1 ME, none Children’s Hospital Muenster, Muenster, Germany; 2Institute of of the ependymomas nor the glioblastoma could be imaged Neuropathology, University Clinics Muenster, Muenster, Germany; 3 using SRI. In selected cases SRI was more sensitive in the Gerhard Domagk Institute of Pathology, University Clinics detection of relapse than conventional imaging by magnetic Muenster, Muenster, Germany; 4Department of Pediatric Hematology and Oncology, University of Iowa, Iowa City, Iowa; 5Department of resonance imaging and computed tomography. Radiology, Division of Nuclear Medicine, University of Iowa, Iowa Conclusions: SRI should be considered in the evalua- City, Iowa; 6Institute for Neuroradiology, University Hospitals tion of residual or recurrent embryonal CNS tumors, espe- 7 Wuerzburg, Wuerzburg, Germany; University Children’s Hospital cially stPNET. The strengths of SRI lie in the differentiation Wuerzburg, Wuerzburg, Germany; 8Clinics and Policlinics for Nuclear Medicine, University of Wuerzburg, Wuerzburg, Germany; of reactive tissue changes versus residual or recurrent tu- and 9Department of Pediatrics, Klinikum Oldenburg gGmbH, mor, the detection of small lesions, and possibly in the Oldenburg, Germany distinction of stPNET from gliomas.

INTRODUCTION ABSTRACT Although neoplasms of glial origin such as glioblastoma Purpose: Although gliomas predominate among central multiforme (GBM) and anaplastic astrocytoma (AA) predomi- nervous system (CNS) neoplasms in adulthood, embryonal nate among malignant central nervous system (CNS) tumors tumors are the most common malignant brain tumors in during adulthood, embryonal neoplasms are most commonly children. Despite novel treatment approaches, including im- encountered in childhood (1). Embryonal tumors comprise a proved radiotherapy and high-dose chemotherapy, survival heterogeneous group of tumors such as ependymoblastoma, rates remain unsatisfactory. The timely diagnosis of residual medulloblastoma (MB), medulloepithelioma (ME), atypical ter- or recurrent embryonal CNS tumors and thus the earliest atoid/rhabdoid tumors (AT/RT), and supratentorial primitive possible time point for intervention is often hampered by neuroectodermal tumors (stPNET; Ref. 2). Because of similar- inaccuracies of conventional imaging techniques. Novel and ities in histology, stPNET have until recently been grouped with refined imaging methodologies are urgently needed. MB (3). It is estimated that stPNET represent 2.5% of all Experimental Design: We have previously demon- childhood brain tumors, affect children 4 weeks to 10 years of strated the use of somatostatin receptor imaging (SRI) in the age, and show a predilection of 2:1 for the male gender (4). In diagnosis of recurrent and residual medulloblastomas. Here, an analysis of the German Brain Tumor trial HIT’91, the 3-year we evaluated somatostatin receptor type 2 (sst ) expression 2 progression free survival rate for 63 children with stPNET was using an antibody in an array of CNS tumors of childhood. 39.1% (5). Treatment for affected patients consists of intensive Eight high-grade gliomas, 4 atypical teratoid/rhabdoid tu- chemotherapy geared at extending the time to ensuing radio- mors, 7 supratentorial primitive neuroectodermal tumors therapy. Prognostic factors in most studies include tumor location (pineal versus nonpineal) and dissemination along the neuraxis (5, 6). The early detection of residual or relapse tumor should Received 8/24/03; revised 1/10/04; accepted 1/22/04. prompt timely intervention and contribute to improved survival. Grant support: Karl Bro¨cker Stiftung, Geseke, Germany. Several investigators have analyzed the value of postoperative The costs of publication of this article were defrayed in part by the surveillance imaging in malignant CNS tumors of childhood, payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to mainly MB (7–9). Although no improvement in survival could indicate this fact. be documented, it was still concluded that surveillance will gain Note: This study was performed in memory of Joachim Ku¨hl, deceased importance with the evolution of new effective treatment mo- September 2003. dalities. Guidelines for the categorization of histological entities Requests for reprints: Michael C. Fru¨hwald, Department of Pediatric Hematology and Oncology, University Children’s Hospital Muenster, for staging and surveillance of childhood brain tumors have Albert-Schweitzer-Strasse 33; 48149 Mu¨nster, Germany. Phone: 0251- been developed (10). The time to detection of a relapse or 83-52491; Fax: 0251-83-47828; E-mail: [email protected]. residual tumor and the sensitivity of imaging modalities may be

Table 1 Results of immunohistochemical staining of malignant central nervous system tumors of childhood

a No. Identifier Diagnosis Age (yrs) Gender sst2 IHC Remark 1 423 stPNET 3 M ϩϩϩb 2 72 stPNET 2/12 F ϩϩϩ 3 2191 stPNET 5 M ϩϩϩϩ 4 1397 stPNET 4 M ϩϩϩ 5 Sc1 stPNET 16/12 M ϩϩ SRI ϩ at relapse 6 Sc3 stPNET 10 F ϩϩϩ SRI ϩ at relapse 7 Sc5 stPNET 4 M ϩϩϩ SRI ϩ at diagnosis and progression 8 549 AT/RT 1/12 F Ϫ 9 226 AT/RT 4 M Ϫ 10 892 AT/RT 7/12 F Ϫ 11 1331 AT/RT 16/12 M ϩϩ 12 1280 AA 2 M Ϫ 13 722 AA 17 F ϩϩϩϩ 14 830 AA 8/12 M ϩϩϩ 15 967 GBM 15 M ϩ 16 1781 GBM 2/12 M ϩ 17 595 GBM 4 M Ϫ 18 1869 GBM 111/12 F ϩϩϩϩ 19 Sc10 GBM 3 M ϩ SRI result questionable 20 694 EP II 16/12 M Ϫ 21 430 EP II 9 F Ϫ 22 728 EP II 15 M Ϫ 23 820 EP II 5 M ϩϩϩϩ 24 699 AE 5 M ϩϩϩϩ 25 951 AE 11 F ϩϩϩϩ 26 837 AE 110/12 M Ϫ 27 Sc11 AE 12 M Ϫ SRI Ϫ at relapse 28 Sc9 cRMS 2 M ϩϩ SRI ϩ at diagnosis and Ϫ postoperation 29 Sc8 ME 8/12 M ϩϩϩϩ SRI ϩ during progression a sst2 IHC, somatostatin receptor type 2 immunohistochemistry; stPNET, supratentorial primitive neuroectodermal tumor; SRI, somatostatin receptor imaging; AT/RT, atypical teratoid/rhabdoid tumor; AA, anaplastic astrocytoma; GBM, glioblastoma multiforme; EP II, ependymoma grade 2; AE, anaplastic ependymoma; cRMS, cerebral rhabdomyosarcoma; ME, medulloepithelioma. b ϪϽ10%; ϩ 11–20%; ϩϩ 21–30%; ϩϩϩ 31–50%; ϩϩϩϩ Ͼ 50% immunopositive cells.

substantially improved as recent methodological advances be- entities were observed (e.g., infants for AT/RT). Twenty-five come Standard of Care (11). patients were male and 9 female. The clinical data on these and We have demonstrated that MB express high levels of the the ones whose tumor were studied by immunohistochemistry

somatostatin receptor subtype sst2 and can be imaged using (IHC) alone are given in Tables 1 and 2. The tumor tissue of 29

scintigraphic somatostatin receptor imaging (SRI) (12). We children was studied by sst2-IHC). Thirteen patients were addi- were able to differentiate nonspecific tissue changes such as tionally studied by SRI. Informed consent was obtained from the inflammation or scar formation from vital tumor tissue. In some legal guardians and if age permitted from the patients as well. cases, SRI proved more sensitive in the detection of residual and All procedures were approved by the respective Institutional relapse tumors than conventional imaging techniques such as Review Boards. computed tomography (CT) and magnetic resonance imaging Tumor Tissue. Tumor material was from patients treated (MRI) and proved a valuable asset in the follow-up of affected at three different institutions (University clinics of Muenster and patients (12). Evaluating somatostatin receptors in MB and Wuerzburg, and University of Iowa). Tissue had been obtained stPNET by reverse transcription-PCR and autoradiography, we at the time of surgery (primary or relapse). Tissue samples were

found that both groups primarily express sst2. Receptor densities processed for routine histopathological examination. Patholog- on stPNET were above those in other tissues that could easily be ical diagnosis was made by local neuropathologists (W. Paulus, studied by SRI (13). Institute of Neuropathology, Muenster, W. Roggendorf, Insti- Until recently, the determination whether a tumor type can tute of Neuropathology, Wuerzburg, and P. Kirby, University of be studied by SRI depended on the availability of sophisticated Iowa) and in the cases from Germany confirmed by the refer- techniques such as radioreceptor-autoradiography. In this study, ence neuropathology panel of the GPOH (University Clinics we evaluated a series of childhood brain tumors using a recently Bonn, Bonn, Germany). None of the patients had previously

developed antibody against sst2. After determination of the been treated with somatostatin or any of the currently available expression status in stPNET, AT/RT, and gliomas in vitro,a analogs of the peptide. series of children were evaluated in vivo using SRI. Immunohistochemical studies were performed on archival paraffin-embedded material from 7 stPNET, 1 ME, 4 AT/RT, 3 MATERIALS AND METHODS AA, 5 GBM, 4 ependymomas grade 2 (EP II), 4 anaplastic Patients. Patients ranged in age from 1 month to 17 ependymomas (AE), and 1 parenchymal cerebral rhabdomyo- years. Typical age distributions for the corresponding tumor sarcoma.

Table 2 Somatostatin receptor imaging in children with malignant central nervous system tumors

a b No. Identifier Dx Age (yrs) Gender Somatostatin receptor imaging Volume sst2 IHC Remark 1 Sc1 stPNET 16/12 M Positive at suspected relapse 30.4 ϩϩ Clinical relapse 2 Sc2 stPNET 8 M Questionable for residual tumor ϳ3 n.d. dod 3 Sc3 stPNET 10 F Positive at relapse 8.34 ϩϩϩ Clinical progression 4 Sc4 stPNET 6 M Positive for residual tumor ϳ3.5 n.d.c Clinical progression 5 Sc5 stPNET 4 M Positive at diagnosis and progression 1.93 ϩϩϩ Clinical progression 6 Sc6 stPNET 2 F Positive at relapse 19.2 n.d. Clinical progression 7 Sc7 stPNET 3 M Positive at relapse 6.33 n.d. Clinical progression 8 Sc8 ME 8/12 M Positive at progression 4.89 ϩϩϩϩ Clinical progression 9 Sc9 cRMS 2 M Positive at diagnosis, negative after 18.5 ϩϩ complete resection 10 Sc10 GBM 3 M Questionable at diagnosis 9 ϩ Clinical progression 11 Sc11 AE 12 M Negative at relapse 3.36 Ϫ Clinical progression 12 Sc12 AE 1 M Negative at relapse 3.6 n.d. Clinical progression 13 Sc13 AE 6 F Questionable at diagnosis 10.92 n.d. MRI ϩ and clinical progression a Volumetry done on magnetic resonance imaging, when only computed tomography information was available ϳϪsigns are given; numbers given in cm3, as calculated for an ellipsoid (axial ϫ sagittal ϫ coronal/2). b sst2 IHC, somatostatin receptor type 2 immunohistochemistry; stPNET, supratentorial primitive neuroectodermal tumor; n.d., not determined; ME, medulloepithelioma; cRMS, cerebral rhabdomyosarcoma; dod, dead of disease; GBM, glioblastoma multiforme; AE, anaplastic ependymoma. c Autoradiography highly positive as published previously (17).

ϫ Somatostatin Receptor sst2-IHC. IHC analysis was ina64 64-matrix. In 4 cases (patients Sc 5, 7, 8, and 12), performed using a monoclonal antibody to sst2. The preparation coregistration of MRI with SRI was performed. All scinti- of the antibody and its application in the study of neuroblasto- graphic imaging results were assessed without prior knowledge mas has been described previously (14). Tumor tissues were of other imaging modalities by two independent observers. SRI fixed in 10% buffered formalin for 24 h, routinely processed, films were classified as true positive and negative or false and paraffin-embedded for histology. IHC was performed on positive or negative in comparison to the results of CT, MRI, ␮ 2- m sections for sst2 at a dilution of 1:2000. The avidin-biotin- and/or tumor histology and CSF evaluation. In all cases, radio- complex method with 3,3Ј-diaminobenzidine as a chromogen tracer uptake in the tumor was compared with uptake in the was used for the detection of the reaction product. skull. Images in which uptake in the tumor was above uptake in To classify staining of tumor cells by the antibody, we the skull (normal bone marrow) were registered as positive. To developed a scoring system. Tumors in which Ͻ10% of the ascertain the potential resolution of SRI versus MRI/CT, vol- tumor cells stained positive were evaluated as negative (Ϫ), umes of suspicious lesions on MRI and CT were determined tumors with 11–20% scored (ϩ), with 21–30% as (ϩϩ), with using the approximation formula for an ellipsoid (coronal ϫ 31–50% as (ϩϩϩ), and all those with Ͼ50% of positive stain- sagittal ϫ axial/2). Infants and children not tolerating the pro- ing as (ϩϩϩϩ). cedure received a mild sedation using oral chloralhydrate 1 h SRI. A total of 13 children was studied by scintigraphy before SRI. Informed consent was obtained from legal guardians and some of them repeatedly (Table 2). For imaging the radio- (in most cases the parents) at least 24 h before the imaging and pharmaceutical In-111-pentetreotide, Octreoscan (Mallinckrodt, associated sedation procedure. Petten, the Netherlands) was injected i.v. using a dose of 3 Mbq/kg body weight. The dose ranged from 30 to 100 Mbq. In-111-pentetreotide is a diethylenetriaminepentaacetic acid RESULTS

(DTPA)-bearing analogue of the octapeptide somatostatin ana- stPNET Consistently Display Expression of sst2 logue octreotide coupled to 111In. Octreotide binds with high The antibody used for our studies has previously been used

affinity the somatostatin receptor subtypes sst2 and sst5 (15). in stably transfected cells, expressing sst2 and, in primary neu- Single-photon emission computed tomography images were ac- roblastomas, expressing the receptor (14). We validated this quired 4 h and in some cases at longer intervals after injection antibody in normal human cerebellum obtained at autopsy. As in employing a Siemens Diacam gamma camera with a medium previous studies, we found abundant staining of the granular cell energy, parallel-hole collimator with energy peaks of 173 and layer in normal human cerebellum (Ref. 16; data not shown). 247 keV (window 20%). In 9 cases, integrity of the blood-brain To screen childhood CNS malignancies justifying evalua- barrier was additionally evaluated using Tc-99m-DTPA as a tion by in vivo SRI, we collected histological entities available tracer. In all these cases, single-photon emission computed in the neuropathology archives of the Institute for Neuropathol- tomography was done in a dual-isotope technique (medium ogy/University Clinics of Muenster. The main focus was em- energy collimator, 173 keV peak for In-111 and 140 keV peak bryonal CNS malignancies but also malignant gliomas. We

for Tc-99m, window 15%). Imaging was performed 4 h after retrieved 29 samples for sst2-IHC (Table 1). Staining patterns injection of the tracer with In-111 and 2 h after injection with were cytoplasmic mostly but also membranous and rarely Tc-99m. A Butterworth filter with a frequency number of 0.6, nuclear (Fig. 1). As expected from our previous studies, all

and an order of six was selected for reconstruction of both scans stPNET stained positive for sst2 (13). Receptor positivity in

Fig. 1 sst2 immunohistochemistry in malignant central nervous system tumors of childhood. An atypical teratoid/rhabdoid tumor (AT/RT), 2 supratentorial primitive neuroectodermal tumor (stPNET), a grade II ependymoma, an anaplastic ependymoma, and a glioblastoma multiforme (GBM)

from children were stained with an antibody against the somatostatin receptor subtype sst2. Abundant receptor protein is detected in the stPNET and the ependymoma II. All others exhibit scattered or only artificial staining.

stPNET was among the highest of all evaluated tumor entities Patient Sc 1. Pineal stPNET in an 18-month-old boy. ϳ ranging between 30 and 80% of immunopositive cells. Of the 4 sst2-IHC: performed on primary tumor material 30% immu- studied AT/RT, only 1 was positive. Among the glial tumors, 3 nopositive cells. SRI: performed after therapy (HIT-SKK ’92 of 4 GBM showed a rather low receptor density with one vincristin, cyclophosphamide, carboplatinum, etoposide, and

exception, which had 80% of cells staining for sst2.Of3AA,1 i.v. methotrexate) when the boy developed signs of irritability had virtually absent immunopositive cells (ϳ5%), whereas 2 and strabism. Comparison SRI versus MRI/CT: SRI detected others were positive with 40 and 80%, respectively. All but 1 of two nodules of high radiotracer uptake. MRI confirmed the 4 EP II were negative for immunostaining. The one with posi- diagnosis, tumor volume 30.4 cm3. The patient has succumbed tive staining had a very high density of immunopositive cells to disease. (90%). Among the AE, 2 of 4 exhibited a high rate of immu- Patient Sc 2. stPNET in an 8-year-old boy. sst2-IHC: no nopositive cells (both 80%), whereas both other tumors had tumor material. SRI: performed after primary surgery. Compar- Ͻ 5% of stained tumor cells. The one studied cerebral rhab- ison SRI versus MRI/CT: SRI without clear radiotracer uptake. domyosarcoma, a very rare entity, had a receptor density of On early postoperative CT, no differentiation of vital tumor ϳ 25% stained tumor cells (Fig. 2). The ME had a density of sst2 versus reactive tissue changes possible, estimated volume 3 cm3 immunopositive cells comparable with stPNET. (CT). On follow-up, MRI residual/recurrence was evident. The patient has succumbed to disease. Patient Sc 3. Ten-year-old girl with a stPNET. sst -IHC: Tumors Expressing High Levels of sst2 Can Be Visualized 2 by SRI on primary tumor 50–60% immunopositive cells. SRI: studied Seven patients with stPNET were evaluated by SRI and at suspected relapse after chemotherapy (HIT ’91, ifosfamide, conventional imaging techniques such as CT and/or MRI. In 4 HD-methotrexate, cisplatinum, and cytarabin) and radiotherapy

cases, information of the in vitro sst2 status of the tumors was (54 Gy). Comparison SRI versus MRI/CT: MRI and SRI imag- obtained in addition to evaluation of the receptor status in vivo ing in excellent concordance demonstrating progressive disease, by SRI. lesion on MRI 8.3 cm3 (Fig. 2).

Fig. 2 Conventional imaging and somatostatin receptor imaging (SRI) in children with embryonal central nervous system malignancies. A, sagittal postcontrast T1-MRI of a 2-year-old female at diagnosis (patient Sc 6). SRI (top row) demonstrates specific uptake of the radiotracer. 99Tc-DTPA demonstrates blood brain barrier disruption. Note higher contrast in the top panel. B, pre- and postsurgical imaging in a 2-year-old boy with a cerebral rhabdomyosarcoma (patient Sc 9). Preoperative T2-MRI shows a parieto-occipital lesion. Post- operative contrast computed tomography reveals no lesion. Preoperative SRI demonstrates abundant tracer uptake, whereas postoperative SRI is negative (bottom panel). C, supratentorial primitive neuroectodermal tumors relapse in a 10-year-old girl. Relapse magnetic resonance imaging (MRI) shows an enhancing right parietal lesion (patient Sc 3). SRI with high tracer uptake (top panel) and BBB disruption as noted by 99Tc-DTPA- study (bottom panel). D, coregistration of MRI and SRI in a 3-year-old with relapse of a supratentorial primitive neuroectodermal tumor (patient Sc 7). Ax- ial and coronal MRIs shows clear enhancement of the tumor that corresponds well with the findings on SRI.

99m Patient Sc 4. Six-year-old boy with stPNET. sst2-IHC: showed a higher uptake compared with the study using Tc- primary tumor tissue previously tested by receptor-autoradiog- DTPA, suggesting receptor-positive tumor tissue. Comparison raphy yielding a high density of receptors (17). SRI: after SRI versus MRI/CT: postoperative CT was read as potential neurosurgical complete resection SRI (48 h after operation) residual disease with a recommendation of close neuroradiologi-

Fig. 3 Somatostatin receptor imaging (SRI) in the early detection of central nervous system tumor relapse. A, postsurgical imaging in a 6-year-old boy (patient Sc 4). Preoperative T2-MRI (left) shows a parieto- occipital lesion. Postoperative contrast computed tomography reveals several nodules that could not be differentiated from postoperative tissue alterations. Postoperative SRI (top panel) with an enhancing lesion indicating vital tumor tissue. 99Tc- DTPA imaging is weaker than SRI but still positive, indicating blood brain barrier disruption. B, top row: sequential T1-MRI images in an 8-month-old male (patient Sc 8). The first three magnetic resonance images (MRIs) were read as postsurgical tissue change. The MRI fur- thest to the right was performed when the patient presented with clinical symptoms and was read as tumor relapse. Middle panel: coregistered SRI shows a gradual increase of tracer uptake suggesting a relapse 2–3 months before confirmation by clinical signs and positive MRI. Bottom panel: comparison of SRI uptake in tumor and skull demonstrates a progression from a ratio of 1.01 tumor/skull to 2.1.

cal monitoring, volume of suspected tissue change ϳ3cm3. SRI and during chemotherapy. Comparison SRI versus MRI/CT: On is clearly positive (Fig. 3A). The patient has died of progressive Tc-DTPA-scintigraphy, a disturbance of the blood-brain barrier disease. is noted, and uptake is significantly higher on SRI (Fig. 2A). SRI Patient Sc 5. Four-year-old boy with a pineal stPNET. is currently used in the follow-up evaluation of this patient, 3 sst2-IHC: performed on primary tumor tissue with a receptor tumor volume 19.2 cm . density of 50%. SRI: studied at diagnosis and during follow-up Patient Sc 7. Three-year-old boy with stPNET. sst2-IHC: 15 months after initial SRI. Comparison SRI versus MRI/CT: no tissue. SRI: imaging performed at suspected relapse. Com- MRI and SRI were coregistered. MRI confirmed the positive parison SRI versus MRI/CT: MRI coregistered with SRI findings of residual tumor after the operation and then progres- showed clear uptake, tumor volume 6.33 cm3 (Fig. 2D). Al- sive disease after chemotherapy and radiotherapy. Furthermore, though single-photon emission computed tomography imaging the suspicion of hippocampal and cerebellar metastases based required mild sedation only, MRI had to be done with full on SRI were confirmed by MRI, nodules with 1.93 cm3. anesthesia.

Patient Sc 6. Two-year-old girl with a stPNET. sst2-IHC: Patient Sc 8. Eight-month-old boy with a ME. Therapy

no tissue. SRI: imaged at suspected relapse after initial surgery consisted of chemotherapy plus confocal radiotherapy. sst2-

IHC: high receptor density with 70–80% of immunopositive developed. Cervera et al. (23) demonstrated the use of using cells. SRI: treatment response was followed by sequential coreg- receptor subtype-specific antibodies by comparing IHC to au- istrations of MRI with SRI. Comparison SRI versus MRI/CT: toradiography in the same tumor samples. We used an antibody SRI 3 days after neurosurgical resection demonstrated minimal generated in the laboratories of one of us and observed mem- uptake of radiotracer. At early follow-up (2 months after sur- branous, cytoplasmic, and nuclear staining patterns (Ref. 14; gery), a suspicious lesion was seen on SRI but not on MRI (MRI Fig. 1). Guyotat et al. (24), in a study on ependymomas, and interpreted as postradiation change, suspicious lesion 4.89 cm3). Reubi et al. (25), in an analysis of neuroblastomas and pheo- Two months later, SRI was read as increasing uptake and chromocytomas, found cytoplasmic staining using sst2 antibod- suspicion of residual disease, whereas MRI was still perceived ies. These authors discovered that autoradiography demon- as therapy associated tissue change. Another month later, the strated membranous receptors in tumors that exhibited a child presented with clumsiness and irritability. MRI confirmed cytoplasmic staining pattern on IHC and that the cytoplasmic the suspicion offered by SRI 4 months earlier, and the diagnosis staining may represent internalized receptors. In fact, as expertly of progression was made. At this time, SRI was highly positive. reviewed by Hofland et al. (26), sst internalization occurs within Volumes of suspected lesions increased from 4.9 to 5.8 to 7.4 to 60 min after binding of ligands. We thus argue that in our 16.3 cm3. MRI was first read as truly positive at 7.4 cm3 (Fig. samples, immunostaining is in part due to membrane-bound and 3B). otherwise due to internalized receptors. We cannot conclude Patient Sc 9. Two-year-old boy with a temporal rhab- whether the receptors are functionally active in the expressing tumors; however, our own studies on MB cell lines and other domyosarcoma. sst2-IHC: up to 30% of positive cells. SRI: studied before and after initial surgery. Comparison SRI versus investigators work on glioma cells demonstrate that intracellular MRI/CT: excellent concordance between preoperative SRI and transduction pathways are activated upon ligand binding (13, MRI (tumor volume, 18.5 cm3). Postoperative imaging by CT 27). and by SRI confirmed the neurosurgeons assessment of a com- The most important finding of our study is the consistent plete neurosurgical resection. expression of sst2 in stPNET and the consecutive ability to Patients Sc 10–13. We furthermore studied four high- image these tumors by SRI. sst2 is expressed in 100% of grade gliomas by SRI. The relapse in 2 ependymomas that had stPNET, whether analyzed by autoradiography, reverse tran- Ͻ scription-PCR, or IHC (13). Receptor densities are above those 5% of sst2-positive tumor cells on IHC did not enhance on SRI (patient Sc 11, 3.36 cm3 and patient Sc 12, 3.6 cm3). One reported for other malignancies, which can easily be imaged by SRI. In only 1 stPNET, SRI could not clearly image the tumor GBM (3 years) with a low density of sst2 on IHC (10%) had questionable radiotracer uptake, and SRI was thus graded as non (patient Sc 2). At suspected residual tumor, neither conventional informative (volume on MRI 9 cm3). The same was true for the imaging by postoperative CT nor SRI could clearly differentiate AE of a 6-year-old girl (patient Sc 13, 10.92 cm3). SRI was between therapy-associated tissue changes and vital tumor tis- questionable whereas MRI and the clinical course confirmed the sue. Follow-up MRI months later proved the presence of vital diagnosis of a progressing tumor. tumor tissue. In 5 cases, SRI was in full accordance with other imaging modalities (patients Sc 1, 3, 5, 6, and 7; Fig. 2). In 1 case (patient Sc 4), SRI was more reliable than postoperative CT DISCUSSION in that it yielded a high radiotracer-uptake (Fig. 3A). Early Despite their rarity, embryonal tumors of the CNS repre- postoperative MRI imaging, preferably within the first 72 h after sent a substantial dilemma in pediatric neuro-oncology. Al- surgery, is well suited for the differentiation of vital enhancing though treatment results have been promising for tumors such as tumor tissue versus blood or scar tissue without the hazards standard risk MB, high-risk MB or AT/RT, and stPNET can associated with the use of radioactivity. However, in facilities rarely be cured. Survival of recurrent stPNET has only been where postoperative MRI may not be available or not possible reported in high-dose chemotherapy trials (18, 19). Because of because of other causes, SRI may be available. It is not so much their histological resemblance, stPNET and MB were until re- the situation of residual tumor, however, where the strength of cently differentiated by their localization (4). Molecular studies the methodology is seated. In patient Sc 8 for instance, coreg- have recently demonstrated differences between MB and stP- istration of MRI along with SRI showed advantages of SRI (Fig. NET. For instance, genetic and epigenetic lesions of the short 3B). Although MRI at 4 months after resection was read as arm of chromosome 17 (17p), frequently detected in MB, are postradiation changes, SRI was strongly suspicious for recurrent rare in stPNET (20, 21). Only sparse data have appeared on tumor. At 5 months, SRI yielded an even stronger positive stPNET and mostly in conjunction with MB. This has mainly signal, and MRI confirmed the diagnosis as well as the clinical been due to a less than clear differentiation between tumor types course with the onset of symptoms such as irritability and ataxia. and to the rarity of stPNET. Calculating from a CCG study Thus, in addition to a perfect correlation between receptor stPNET occur in ϳ1 in 10 million children (6). It is thus not expression in vitro and imaging by SRI in vivo, there was an surprising that among the cases for our study only 6 could be excellent correlation between conventional imaging techniques evaluated by IHC and seven by SRI. and SRI for stPNET. As outlined, these correlations may be very The presence of sst receptors on tumor tissue has until useful in the diagnosis of early recurrent or residual tumors that recently been assessed by the use of radioactively labeled pep- cannot be imaged by other modalities. tide analogues (22). A major drawback of this approach is the The volumes of the suspected lesions ranged in size from requirement for fresh tissue and a technically demanding pro- 1.93 to 30.4 cm3. However, whereas the smallest lesion (patient cedure. As a consequence, antibodies for each sst have been Sc 5; 1.93 cm3) could be visualized by SRI, larger lesions could

90 not when sst2 receptor densities were too low (patient Sc 10; brachytherapy Merlo et al. (33) used a Y-labeled somatostatin 3.36 cm3). Furthermore, in the few cases studied by SRI and analogue to control the growth of a series of gliomas. These Tc-DTPA to assess blood-brain barrier permeability, we found authors found this approach not only safe but also efficient in an excellent correlation. In all cases where Tc-DTPA showed the locoregional control of the progression in low to intermedi- disruption of the blood brain barrier, SRI was positive and ate grade gliomas. Similar approaches may well be taken for showed a larger region of enhancement. SRI enhancement is stPNET. highly specific for sst2-bearing tissue and not likely due to Several questions not answered by our rather small cohort nonspecific extravasation into surrounding brain edema. This need to be addressed in future studies on more patients. Al- has previously been demonstrated for gliomas and meningeo- though we were able to ascertain that a positive in vitro status of mas but also for MB (12, 28, 29). Meewes et al. (29), for sst2 expression correlates well with the imaging of tumors at instance, showed less intense uptake in TC-DTPA images than diagnosis and relapse as well as in residual tumors and metas- in SRI. Furthermore, in patients Sc 1, 3, 4, 5, 8, and 9, we found tases (patients Sc 7 and 8), we could not establish the limit of a perfect correlation between in vitro receptor status and in vivo resolution of SRI. However, as mentioned above, studies on imaging using SRI. meningeomas and gliomas estimate that lesions need to be at 3 As we were able to image lesions as small as 1.93 cm , SRI least 1 cm3 in size to be reliably imaged (28, 29). Second, may help to image sst2-positive stPNETs in regions with high patients with suspected relapse may in the future undergo an background enhancement on MRI (i.e., choroid plexus, inflam- array of imaging procedures such as diffusion weighted MRI, matory tissue). Specific uptake in sst2-bearing tissues should MRI spectroscopy, and nuclear medicine technologies (e.g., prove more pronounced in the sst -positive tissues than the 2 SRI). The advantage of our approach may be that tumors can be negative tissues or those with low activity. tested in vitro to ascertain that in vivo imaging by SRI is Currently, the differential diagnosis between malignant worthwhile. In our study on MB, we tested if imaging after 72 h glioma and stPNET is close to impossible by MRI or CT so that yields any additional information. For the cohort of stPNET, we surgical sampling is performed. As we and others have shown, chose to abandon this route because no additional information malignant gliomas are almost always negative on SRI. Thus, a but a gradual decay could be noted and because most children supratentorially located tumor with MRI and CT features char- are in a critical postoperation phase not warranting any unnec- acteristic of a malignant process (perifocal edema, mixed en- essary procedures. Follow-up of our patients has not been long hancement), which enhances on SRI, is very likely a stPNET. This may be of major importance as novel therapeutic strategies, enough to warrant investigation as to whether early diagnosis including neoadjuvant chemotherapy, are used to make the has a real clinical impact on survival or more realistically tumor amenable to a complete neurosurgical resection. quality of life. Up to 40% of supratentorially located PNET exhibit me- Some of the brain tumors evaluated in our study have tastases at diagnosis, and an even higher number cannot be previously not been assessed for sst2 expression. AT/RT have reliably resected (5). Furthermore, conventional treatment strat- gained major attention as a clinical as well as a biological egies have been rather disappointing with low survival rates for enigma (34). In three of our samples, staining was faint to primary tumors and numbers close to zero for recurrent tumors. absent. In contrast, another specimen had 30% immunopositive The presence of peptide receptors that can be targeted by a cells mainly of the neuroectodermal component. It remains radiolabeled ligand indicate therapeutic alternatives. In fact, speculative whether tumors that showed no sst2 staining repre- somatostatin analogues might be used i.v. because stPNET sent a subvariant of AT/RT. almost inevitably demonstrate a disrupted blood-tumor barrier. The studied rhabdomyosarcoma had an intermediate recep- As many of these tumors require access to the CSF via an tor density of ϳ30% immunopositive cells. The studied ME implanted catheter direct application into the CSF is possible. showed an even stronger expression of the sst2 protein. SRI was Although therapy with somatostatin and its congeners has been highly positive in both tumor entities. Both tumor types are disappointing in most solid cancers, Kouroumalis et al. (30) thought to be derived from primitive neuroepithelium. The

raised some hope. These authors treated patients with hepato- rather abundant expression of sst2 on these and on stPNET and cellular carcinoma and found additive beneficial effects when MB may be the basis for a unifying concept. octreotide was combined with conventional chemotherapy. With The data on sst expression in glial tumors are conflicting. the development of subtype-specific sst ligands but also of Although Reubi et al. (35) detected sst in only 1 of 20 GBM by ligands with high affinity for all five sst subtypes, it may be autoradiography, others (36, 37) found specific binding of so- possible to influence the proliferation of solid tumors such as matostatin analogues on GBM. Lamszus et al. (38) and Feindt et

stPNET (15). Some researchers have linked somatostatin ana- al. (27) detected sst2 mRNA in GBM and concluded that sst2 logues to cytotoxic agents. Kiaris et al. (31) used a doxorubicin expression is not abundant but present on GBM. Using subtype-

derivative linked to the somatostatin analogue RC-160 devel- specific antibodies for sst2 and sst3, as well as autoradiography oped by Nagy et al. (31) to efficiently suppress the growth of and reverse transcription-PCR, Cervera et al. (23) determined glioma xenograft tumors in mice. Very high potential lies in the microglia and proliferating vessels as the main source for sst approach of peptide receptor radionuclide therapy. Somatostatin receptors in GBM. In our analysis, all of the examined GBM that can be labeled with either high energy 111In or the ␤-emitter showed some degree of sst immunopositivity. This contradic- 90Yorthe␣ emitter 177Lu have been developed. Vaidyanathan tory finding may in part be reconciled by the different nature of et al. (32) used novel iodinated somatostatin analogues that GBM in children and adults, e.g., TP53 mutations, which are could be targeted to MB xenografts. In a pilot study using often found in adult GBM, are sparse in childhood GBM (39).

Among the evaluated AA we, as most other authors, found a recurrent brain tumors in asymptomatic children. J Neuro-Oncol 2001; varying pattern. 53:33–8. 10. Kramer ED, Vezina LG, Packer RJ, et al. Staging and surveillance We found sst2 expression on most of the ependymomas; as had Guyotat before (24). In our specimen, which differed from of children with central nervous system neoplasms: recommendations of the Neurology and Tumor Imaging Committees of the Children’s Can- the ones studied by Guyotat et al. (24) in that they were derived cer Group. Pediatr Neurosurg 1994;20:254–62. from children (1–15 years), immunostaining for sst2 was absent Ͼ 11. Erdem E, Zimmerman RA, Haselgrove JC, Bilaniuk LT, Hunter JV. in 50% (3 of 4 EP II and 2 of 4 AE). However, in 3 tumors (2 Diffusion-weighted imaging and fluid attenuated inversion recovery AE and 1 EPII), we found rather dense receptor distribution imaging in the evaluation of primitive neuroectodermal tumors. Neuro- (80% in 2 AE and 90% in an EP II). From our limited numbers, radiology 2001;43:927–33.

AE showed higher sst2 expression than did lower grade tumors. 12. Mu¨ller HL, Fru¨hwald MC, Scheubeck M, et al. A possible role for This is confirmed by Guyotat et al. (24) in adult ependymomas, somatostatin receptor scintigraphy in the diagnosis and follow-up of children with medulloblastoma. J Neuro-Oncol 1998;38:27–40. where AE had higher sst2 levels than EP II, which had again higher levels than EP I. The question whether ependymomas of 13. Fru¨hwald MC, O’Dorisio MS, Pietsch T, Reubi JC. High expression of somatostatin receptor subtype 2 (sst2) in medulloblastoma: implica- childhood express sst2 protein will need to be evaluated in larger tions for diagnosis and therapy. Pediatr Res 1999;45:697–708. cooperative studies. In contrast to Guyotat, who was able to 14. Albers AR, O’Dorisio MS, Balster DA, et al. Somatostatin receptor image all ependymomas, SRI was not helpful in the diagnosis of gene expression in neuroblastoma. Regul Pept 2000;88:61–73. ependymoma recurrence in our patients. Our data are supported 15. Lamberts SW, de Herder WW, Hofland LJ. Somatostatin analogs in by other investigators, who could not image ependymomas by the diagnosis and treatment of cancer. Trends Endocrinol Metab 2002; SRI (27, 40). 13:451–7. In conclusion, we have demonstrated that stPNET and 16. Schindler M, Holloway S, Humphrey PP, et al. Localization of the some other rare neoplasias of the CNS during childhood express somatostatin sst2(a) receptor in human cerebral cortex, hippocampus and cerebellum. Neuroreport 1998;9:521–5. levels of sst2 that warrant scintigraphic imaging, if the blood- brain barrier is disrupted. In selected cases, SRI was more 17. Fru¨hwald MC, O’Dorisio MS, Pietsch T, Reubi JC. Receptors for somatostatin (sms) and vasoactive intestinal peptide (VIP) in medullo- sensitive for the detection of small lesions than MRI and CT. blastoma as diagnostic and therapeutic targets. Neuro-Oncol 1999;1:19. SRI thus holds great promise as an asset in the imaging of 18. Graham ML, Herndon JE II, Casey JR, et al. High-dose chemother- residual and/or recurrent stPNET. Our cohort is not yet large apy with autologous stem-cell rescue in patients with recurrent and enough to precisely determine sensitivity and specificity of SRI high-risk pediatric brain tumors. J Clin Oncol 1997;15:1814–23. in children with stPNET. More patients with this rare disease 19. Gururangan S, McLaughlin C, Quinn J, et al. High-dose chemo- can only be studied in a multi-institutional fashion. therapy with autologous stem-cell rescue in children and adults with newly diagnosed pineoblastomas. J Clin Oncol 2003;21:2187–91. 20. Russo C, Pellarin M, Tingby O, et al. Comparative genomic hy- ACKNOWLEDGMENTS bridization in patients with supratentorial and infratentorial primitive We thank Birgit Kallinger and Beta Hillmann for their expert neuroectodermal tumors. Cancer (Phila) 1999;86:331–9. technical assistance. 21. Fru¨hwald MC, O’Dorisio MS, Dai Z, et al. Aberrant hypermethy- lation of the major breakpoint cluster region in 17p11.2 in medulloblastomas but not supratentorial PNETs. Genes Chromosomes Cancer 2001; REFERENCES 30:38–47. 1. Gilbertson R. Paediatric embryonic brain tumours. Biological and 22. Reubi JC, Waser B, Schaer JC, Laissue JA. Somatostatin receptor clinical relevance of molecular genetic abnormalities. Eur J Cancer sst1-sst5 expression in normal and neoplastic human tissues using 2002;38:675–85. receptor autoradiography with subtype-selective ligands. Eur J Nucl 2. Kleihues P, Louis DN, Scheithauer BW, et al. The WHO classifica- Med 2001;28:836–46. tion of tumors of the nervous system. J Neuropathol Exp Neurol 2002; 23. Cervera P, Videau C, Viollet C, et al. Comparison of somatostatin 61:215–25. receptor expression in human gliomas and medulloblastomas. J Neu- 3. Rorke LB. The cerebellar medulloblastoma and its relationship to roendocrinol 2002;14:458–71. primitive neuroectodermal tumors. J Neuropathol & Exp Neurol 1983; 24. Guyotat J, Champier J, Jouvet A, et al. Differential expression of 42:1–15. somatostatin receptors in ependymoma: implications for diagnosis. Int J 4. Rorke LB, Trojanowski JQ, Lee VM, et al. Primitive neuroectoder- Cancer 2001;95:144–51. mal tumors of the central nervous system. Brain Pathol 1997;7:765–84. 25. Reubi JC, Waser B, Liu Q, Laissue JA, Schonbrunn A. Subcellular 5. Timmermann B, Kortmann RD, Kuhl J, et al. Role of radiotherapy in distribution of somatostatin sst2A receptors in human tumors of the the treatment of supratentorial primitive neuroectodermal tumors in nervous and neuroendocrine systems: membranous versus intracellular childhood: results of the prospective German brain tumor trials HIT location. J Clin Endocrinol Metab 2000;85:3882–91. 88/89 and 91. J Clin Oncol 2002;20:842–9. 26. Hofland LJ, Lamberts SW. The pathophysiological consequences of 6. Albright AL, Wisoff JH, Zeltzer P, et al. Prognostic factors in somatostatin receptor internalization and resistance. Endocr Rev 2003; children with supratentorial (nonpineal) primitive neuroectodermal tu- 24:28–47. mors. A neurosurgical perspective from the Children’s Cancer Group. 27. Feindt J, Becker I, Blomer U, et al. Expression of somatostatin Pediatr Neurosurg 1995;22:1–7. receptor subtypes in cultured astrocytes and gliomas. J Neurochem 7. Torres CF, Rebsamen S, Silber JH, et al. Surveillance scanning of 1995;65:1997–2005. children with medulloblastoma. N Engl J Med 1994;330:892–5. 28. Haldemann AR, Rosler H, Barth A, et al. Somatostatin receptor 8. Steinbok P, Hentschel S, Cochrane DD, Kestle JR. Value of postop- scintigraphy in central nervous system tumors: role of blood-brain erative surveillance imaging in the management of children with some barrier permeability. J Nucl Med 1995;36:403–10. common brain tumors. J Neurosurg 1996;84:726–32. 29. Meewes C, Bohuslavizki KH, Krisch B, et al. Molecular biologic 9. Korones DN, Butterfield R, Meyers SP, Constine LS. The role of and scintigraphic analyses of somatostatin receptor-negative meningio- surveillance magnetic resonance imaging (MRI) scanning in detecting mas. J Nucl Med 2001;42:1338–45.

30. Kouroumalis E, Skordilis P, Thermos K, et al. Treatment of hepa- mors of the human central nervous system. Cancer Res 1987;47:5758– tocellular carcinoma with octreotide: a randomised controlled study. Gut 64. 1998;42:442–7. 36. Pinski J, Schally AV, Halmos G, Szepeshazi K, Groot K. Soma- 31. Kiaris H, Schally AV, Nagy A, et al. Regression of U-87 MG tostatin analogues and bombesin/gastrin-releasing peptide antagonist human glioblastomas in nude mice after treatment with a cytotoxic RC-3095 inhibit the growth of human glioblastomas in vitro and in vivo. somatostatin analog AN-238. Clin Cancer Res 2000;6:709–17. Cancer Res 1994;54:5895–901. 32. Vaidyanathan G, Friedman HS, Affleck DJ, et al. Specific and 37. Luyken C, Hildebrandt G, Scheidhauer K, et al. 111Indium (DTPA- high-level targeting of radiolabeled octreotide analogues to human octreotide) scintigraphy in patients with cerebral gliomas. Acta Neuro- medulloblastoma xenografts. Clin Cancer Res 2003;9:1868–76. chir 1994;127:60–4. 33. Merlo A, Hausmann O, Wasner M, et al. Locoregional regulatory 38. Lamszus K, Meyerhof W, Westphal M. Somatostatin and soma- peptide receptor targeting with the diffusible somatostatin analogue tostatin receptors in the diagnosis and treatment of gliomas. J Neuro- 90Y-labeled DOTA0-D-Phe1-Tyr3-octreotide (DOTATOC): a pilot Oncol 1997;35:353–64. study in human gliomas. Clin Cancer Res 1999;5:1025–33. 39. Sure U, Ruedi D, Tachibana O, et al. Determination of p53 muta- 34. Packer RJ, Biegel JA, Blaney S, et al. Atypical teratoid/rhabdoid tions, EGFR overexpression, and loss of p16 expression in pediatric tumor of the central nervous system: report on workshop. J Pediatr glioblastomas. J Neuropathol & Exp Neurol 1997;56:782–9. Hematol Oncol 2002;24:337–42. 40. Maini CL, Sciuto R, Tofani A, et al. Somatostatin receptor imaging 35. Reubi JC, Lang W, Maurer R, Koper JW, Lamberts SW. Distribu- in CNS tumours using 111In-octreotide. Nucl Med Commun 1995;16: tion and biochemical characterization of somatostatin receptors in tu- 756–66.

Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2004 American Association for Cancer Research. Somatostatin Receptor Subtype 2 Is Expressed by Supratentorial Primitive Neuroectodermal Tumors of Childhood and Can Be Targeted for Somatostatin Receptor Imaging

Michael C. Frühwald, Christian H. Rickert, M. Sue O'Dorisio, et al.

Clin Cancer Res 2004;10:2997-3006.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/10/9/2997

Cited articles This article cites 38 articles, 11 of which you can access for free at: http://clincancerres.aacrjournals.org/content/10/9/2997.full#ref-list-1

Citing articles This article has been cited by 3 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/10/9/2997.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/10/9/2997. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.