[1] 16w In Vivo Metabolism of Childhood Posterior Fossa Tumors and Primitive Neuroectodermal Tumors before and after Treatment

[1] 287w Background. Despite an increasing interest in the clinical application of positron emission tomography (PET) in tumors of the adult patient as a diagnostic and prognostic tool, only a few studies have been concerned with the usefulness of PET with [18F]2-deoxy-2-fluoro-Dglucose (FDG) in childhood tumors.Methods. Fifteen children and young adults (0.5-26.0 years of age) with histologically confirmed brain tumors were studied with PET and FDG. Seven children with medulloblastoma (n = 5) or primitive neuroectodermal tumor (PNET) (n = 2) underwent repeated PET studies during their therapy. The other eight children with medulloblastoma (n = 5) or astrocytoma (n = 3) were studied only once before initiation of treatment. A close clinical follow-up was performed in every case.Results. Comparison of local glucose metabolic rates obtained in the various tumor lesions revealed that the mean rates found in medulloblastomas (mean glucose metabolic rate, 42.8 ? 14.03 pmo1/100 g/min) were twice as high as the rates measured in either PNET (17.3 k 4.5 pmo1/100g/min) or infratentorial gliomas (21.8 f 4.2 pmo1/100 g/min). The high metabolism of medulloblastomas enabled the observer to identify the tumor more easily and to clearly separate it from the surrounding unaffected brain tissue. In the seven patients with follow-up PET studies during therapy, decreasing or increasing ratios of tumor-to-white matter metabolic rate were not only commensurate with neuroradiologically defined tumor reduction or growth, but also corresponded to the duration of initial clinical improvement. Conclusions. These preliminary results suggest that PET with FDG may be a useful tool to evaluate metabolic activity of pediatric brain tumors over time and to assess response to treatment. Cancer 1993; 721394-403.From the Max-Planck-Institut fur Neurologische Forschung, the *Kinderklinik der Universitat zu Koln, and the thstitut fur Pathologie der Universitat zu Koln,

[1] 289w A radial artery and cutaneous vein were cannulated for withdrawal of blood samples and for injection of the radioisotope, respectively. The subjects were positioned supine in the gantry of a four-ring positron tomograph (Scanditronix PC 384, Sweden). A total of 14 partially overlapping, transaxial images of brain glucose metabo- ) Site of primary Tumor after Tumor location at Patient Age (yr) Sex Histology tumor* resection* PET scan 8 9.4 M Medulloblastoma Edge ventricle IV Recurrent after Brain stem total gross removal 9 11.9 M Medulloblastoma Posterior fossa Total gross removal Right occipital cortex tumor, of posterior fossa supratentorial tumor, metastases supraten torial metastases 10 9.6 M Medulloblastoma Posterior fossa Supratentorial Left frontobasal 11 15.9 M Medulloblastoma Left cerebellum Recurrent after Left cerebellum tumor metastases cortex total gross removal 12 5.5 M Medulloblastoma Right cerebellum Recurrent after Left cerebellum, right total gross cerebellum, right removal cerebellar peduncle 13 7.9 M Astrocytoma grade I Posterior fossa Primary, Pons, right up to stereotactic cerebellum, left supratentorial biopsy thalamus location 14 11.5 F Pilocytic astrocytoma Posterior fossa Primary, Right cerebellar stereotactic peduncle biopsy stereotactic peduncle, pons biopsy 15 9.1 M Pilocytic astrocytoma Posterior fossa Primary, Right cerebellar PET positron-emission tomography. * Shown by computed tomography or magnetic resonance imaging scan. lismI6 were recorded in parallel with the canthomeatal line. A dynamic series of PET data and approximately 10 arterial blood samples were obtained over 50 minutes following intravenous injection of 100 pCi FDG/ kg of body weight. In-plane resolution was 7.8 mm at a slice thickness of 11 mm. The older children were studied in a resting state with their eyes closed and ears unplugged in a darkened room with low-ambient noise. 2. The treatment referred to is the HIT-89 trial (Fig. 1).

[2] 196w Clinical and histopathologic data as well as a description of the various tumor locations obtained in the 15 children and young adults under study are summarized in Table 1. Repeated PET measurements were performed in seven patients (Group l), and eight children underwent a single PET study only (Group 2). In Group 1, five brain tumors (one with three lesions) were diagnosed as medulloblastomas, thus a total of seven lesions were studied. Three lesions were found in the two children diagnosed with PNET. In Group 2, we examined five cases of medulloblastoma with a total of seven lesions and three first grade astrocytomas with a total of six lesions. In both groups together, four patients presented with primary tumors, and the other patients presented with tumor recurrence. The glucose metabolic rates measured in the tumor lesions are summarized in Tables 2 and 4. Given are absolute rates of tumor glucose metabolism as well as ratios of tumor-to-white-matter metabolic rates. Tables 2 and 4 also summarize at what point during therapy, in days after last treatment for Group 1 (Table 2) or in months after last therapy for Group 2 (Table 4), a PET scan was performed.

[3] 168w A comparison of tumor metabolism observed in medulloblastoma, PNET, and glioma lesions was achieved by averaging over the tumor metabolic rates obtained for each lesion in the initial PET scan. The average rate obtained in primary medulloblastoma lesions or metastases (14 lesions; mean metabolic rate, 42.8 2 14.0 pmol/100 g/min; range, 22-74 pmo1/100 g/min) was typically much higher than the rates found for PNET (n = 3; 17.3 f 4.51 pmo1/100 g/min; range, 13-22 pmol/ 100 g/min) or glioma lesions (n = 6; 21.8 f 4.22 pmol/ 100 g/min; range, 16-26 pmo1/100 g/min). Metabolism was also high in the four supratentorial medulloblastoma metastases included in that group (Figs. 234). In patients with pentobarbital sedation before the PET study, glucose metabolism in the tumor lesions was similar to the rates found in the other cases. Although the number of lesions studied is small, the tendency toward high metabolism in medulloblastomas was obvious in every case. This property enabled the observer to clearly separate tumor mass from the surrounding tissue.

[4] 221w In Group 1 patients, the lesion with the highest metabolic rate for glucose in the initial scan was used to demonstrate the change in tumor-to-white-matter metabolic ratio during treatment and follow-up scans (Table 2). When comparing the months of clinical relief and the changes demonstrated in CT or MRI scan of Group 1 with the changes in the ratios calculated for tumor-to-white metabolic rates (Table 2), there seems to be a tendency toward longer remission periods in children with the more important decreases in tumor metabolism during treatment. Smaller changes were found in patients with shorter survival. We encountered one case of medulloblastoma (Patient 7) where, despite aggressive chemotherapeutic treatment before the first PET scan and radiation therapy before the second PET study, tumor size remained unaffected and tumor metabolism even increased (Fig. 3). In all other cases of medulloblastomas, we observed a decrease in tumor size and metabolism (Fig. 2), although the effect was very small (11%) in one of the PNET (Patient 6). Patients A.M.A. and M.E.R. showed positive initial responses to chemotherapy, and tumor metabolic rates decreased markedly. Both patients, however, presented with supratentorial metastases at the time of treatment initiation (Patient 1) or developed additional metastases during the months following completion of therapy (Patient 2), as shown in Figure 3. The patients died of their cerebral metastases.

[5] 39w Global gray matter metabolism was unchanged in children with follow-up PET scans after treatment when comparing the initial scan (mean global rate, 27.3 f 6.26 pmol/100 g/min) to the last scan (mean global rate, 25.2 -t 5.00 pmo1/100 g/min).

[6] 249w Tracer kinetic analysis in three patients revealed that the high values for FDG uptake into medulloblastoma lesions are due to increased FDG phosphorylation, whereas FDG transport into the lesion is normal (Fig. 5). play an important role in the in vivo assessment of brain tumors." However, only very few studies have been concerned with the possible value of PET in childhood tumors. '',15 One of the main differences between childhood tumors and adult tumors lies in the nature of the child's nervous system. During the first 2 years of life and particularly during the first 6 months, the brain is involved in a continuing growth spurt. Its increase in size is due to both progressive myelination and to arborization of neuronal dendrites. Although the infant brain is more plastic in its response to surgical and accidental trauma, the effects of hypoxia, radiation therapy, and chemotherapy on vital developmental processes may produce long-term interference with intellectual, motor, and hormonal It has been evident for many years that a variety of cytotoxic agents as well as radiation therapy have substantially improved cures in children with medulloblastomas and PNET.2-6 In the context of a considerable chance of cure, there seems no doubt that the possible hazards of treatment have to be accepted. Nevertheless, it is important to develop methods that have the potential to provide clinicians with information on tumor progression and treatment effects at a time where therapy can still be modified, which would also, in selected cases, reduce treatmentrelated sequelae.

[1] 101w We studied a total of 10 children with medulloblastoma, 2 children with PNET, and 3 children with posterior fossa gliomas. Only four of our patients were de novo diagnosis of posterior fossa tumors, whereas all other cases were either recurrent tumors and/or metastases. Tumor recurrence in the posterior fossa is a frequent cause of clinical relapse. Furthermore, metastatic deposits at remote sites often invade the adjacent neural tissues diffusely, either by direct extension or by way of the perivascular sheaths of perforating blood vessels.*' Due to their rapid development and size, these deposits may be responsible for neurologic signs and symptoms.

[2] 139w Sedation with pentobarbital before each PET study was necessary in three children with serial PET scans during therapy. It has been suggested recently that barbiturates reduce neuronal metabolism by blocking synaptic a ~t i v i t y . ~~, ~~ Theodorez4 demonstrated that phenobarbital causes a mean reduction over all regions (white and gray matter) of 37% in patients with epilepsy. Gliomas, however, have been shown to resist suppression of glucose metabolism during barbiturate coma to some extent, which was explained by lack of neuronal activity in tumor lesions.z4 Based on these findings, we assumed that barbiturate sedation in our three patients would have had a similar global effect on overall glucose consumption in normal tissue in each study and a neglectable effect on tumor metabolism. Therefore, a comparison between scans acquired in the same patient appeared reasonable.

[3] 189w When comparing the mean tumor glucose metabolic rate in the different histologic groups, we were surprised to observe a rate in medulloblastoma lesions more than twice as high as the rates found in three PNET lesions. Although only a small number of lesions were studied, the observation of high rates in medulloblastoma lesions was a consistent finding. The category PNET was coined by Hart and EarleZ5 to describe tumors occurring in the cerebral hemispheres that resemble medulloblastoma histologically, but arise in an area distant from the characteristic site of origin of the medulloblastoma. They can grow at any supratentorial site ~ and are often extremely large and radiologically bizarre. To our knowledge, no histologic differences between medulloblastoma and PNET have been described that could account for a difference in metabolic behavior. Given the small sample size (three PNET lesions), this finding needs further amplification. Analysis of tracer kinetics in three cases of medulloblastoma demonstrated that the high-metabolic rate was due to increased FDG phosphorylation (k3), whereas FDG transport from blood to tissue (K,) was unchanged. This behavior has also been shown to be characteristic for malignant gliomas in adults.z6

[4] 136w The high metabolic rate found in medulloblastoma lesions, however, seems to be useful in two different ways. Due to the bigger contrast between tumor and surrounding tissue, the lesions can be distinguished more easily from unaffected brain and possibly also from late radionecrosis, which has been shown to have a low-metabolic rate in adult patient^.",'^ Furthermore, larger study may also be able to demonstrate that FDG PET offers limited help in the differentiation between brain tumor types, medulloblastoma lesions having the highest metabolic rate of the most common posterior fossa tumors. Despite the aggressive chemotherapy, no change in global neuronal metabolism could be observed in the follow-up study following therapy. Given the relatively short interval between completion of treatment and follow-up PET performed in the majority of children under study, late treatment-related sequelae cannot be excluded.

[5] 109w Seven of our young patients underwent repeated PET studies. We compared the decrease in tumor metabolic rate obtained during treatment with the duration of initial clinical improvement. A report published pre-viouslyZ7 has demonstrated an increase in tumor metabolism 24 hours after chemotherapy and a decrease in baseline thereafter, which is possibly due to tissue metabolic abnormalities. To avoid misinterpretation of tumor activity, we waited for at least 7 days after completion of chemotherapy before performing the individual PET scans. There was one exception: in Patient 1, the initial PET scan was done 3 days after element I, which is an oral therapy during week 1 of the HIT-89 protocol.

[6] 168w Our data suggest that the more marked the decrease in tumor metabolism following chemotherapy was, the longer the period of initial clinical improvement lasted. Although the decrease in tumor metabolism seems to be an indicator of therapeutic success, the cases presented here also prove that other features, such as histologic tumor type, tumor location, and spread as well as age, appear to have an important impact on the patient's outcome. Attempts have been made recently to identify factors of statistical significance, including size, tumor location, presence and absence of metastatic spread, and histopathologic features that collectively might have a bearing on prognosis.'* Although the results were suggestive, it was not possible to confirm the prognostic value of some of these factor^.'^ It has also been shown that children younger than 3 years of age at diagnosis have a lower survival rate. Whether this is due to less aggressive treatment, the tendency for younger children to have disseminated disease at diagnosis or differences of tumor biology is unknown. 2,4,3Q

[7] 215w In our series, it was evident that the patients with the least clinical improvement also showed the smallest decrease in tumor metabolism. After treatment, we even observed an increase in tumor metabolic rate in one of those patients (Patient 7). This lack of therapeutic success may be attributed to the histologic type of tumor encountered in those children as well as their age. Of the two patients with PNET (Patients 5 and 6), one tumor was inoperable (Patient 6) and both were the youngest children under study. One of the patients with medulloblastoma presented with a tumor including large parts of a medullomyoblastoma (Patient 7), which is uncommon and known to be highly malignant. The two patients with the best response to treatment in terms of decline in tumor metabolism lived a normal life for 8 and 10 months, respectively (Patients 1 and 2). Both patients died of their supratentorial metastases that occurred after completion of therapy. Two of our patients with a good tumor metabolic response to treatment (Patients 3 and 4) have lived in complete clinical remission for 17 months. One of these patients suffered of a histologically confirmed desmoplastic medulloblastoma. The desmoplastic medulloblastomas in adolescents have often been considered to carry a more favorable prognosis than those in the midline in children.31

[8] 54w Long-term follow-up of patients who had childhood acute lymphocytic leucemia have shown alterations in gray matter metabolism associated with cranial radiation and ~hemotherapy.~' In the seven patients presented here, no alterations of gray matter metabolism could be observed following chemothera-peutic and/or radiation treatment, although late effects can only be excluded with later follow-up studies.

[9] 97w Our series is too small to define the specific usefulness of PET and FDG in childhood tumors, and the present study needs further amplification. However, our findings suggest that future studies of tumor glucose metabolism in children with posterior fossa tumors hold promise for a specific role of PET in evaluating early treatment response and, in selected cases, to obviate the need for further chemotherapy and/or radiation therapy. This would be a further advance toward maximal reduction of treatment-related sequelae, which is an import and frequent issue arising in the management of pediatric patients with brain tumors.

[1] 296w Fifteen children and young adults (10 male, 5 female, ages 0.5-26 years) with a brain tumor were studied with PET and FDG. When seen by the University of Cologne pediatric services, 11 patients presented with residual or recurrent brain tumor or metastases after total or subtotal resection and 4 patients presented with a primary tumor (Table 1). The diagnosis of brain tumor was confirmed histologically and included 10 patients with medulloblastoma (8 medulloblastomas of the classical type, 1 of the desmoplastic type, and 1 medullomyoblastoma), 2 patients with PNET, and 3 patients with astrocytomas (Table 1). Neuroradiologic studies (computed tomography [CT] and magnetic resonance imaging [MRI]) performed before every PET study were available in each case. Clinical data of the two groups of patients under study are summarized in Table 1. The first group comprised seven young patients who underwent repeated FDG PET studies at different stages during their therapy (Group 1, Tables 1 and 2). All patients in Group 1 received high-dose polychemotherapy according to the HIT-89 trial (Fig. 1) of the German Society of Pediatric Oncology. Of the eight patients studied only once (Group 2, Tables 3 and 4), one was diagnosed de novo and the remaining tumors were recurrent. Surgery, if performed, had been completed before the first PET study in both groups. Written informed consent was obtained from the parents or patients of full age. All patients were followed by regular clinical examinations after their last PET study. Group 1 children were observed very closely over the months following therapy and their clinical development was evaluated. The patient's condition was described as improved when neurologic impairment decreased and when life normalized in terms of eating, playing, and attending kindergarten or school (Table 2). Lack of prominent change was called a stable condition.

[2] 268w Table 2. Glucose Metabolic Rates and Clinical Outcome (Group 1) Tumor glucose Clinical course after PET metabolic rate (pmo1/100 g/ Tumor/white and final outcome tumor/white min) lesion matter lesion metabolic ratiot Time since last Initial clinical improvement YO change Patient TherapyandPET scan* A B C A B C PET scan (mo) Tumor Change 1 1st cycle After element l(3) After element I1 (8) After completion (10) Before element I After completion (15) After radiation (330) Before element I Before element I1 Before element I After completion (7) After radiation (46) Before element I After completion (17) Before element I After completion (8) After completion (19) After radiation (50) 2 1st cycle 3 1st cycle 2nd cycle 4 1st cycle 5 1st cycle 6 1st cycle 7 2ndcycle 55 51 74 2.39 2.22 3.20 25 25 26 1.09 1.09 1.13 20 22 22 0.96 1.05 1.10 39 2.14 23 1.00 22 1.03 31 1.85 22 1.06 35 2.17 30 1.86 21 1.30 Died of new C -65.6 metastases (9) 8 10 Died of new A -52.0 metastases (11) Complete clinical A -42.7 remission 17 17 Complete clinical A -40.1 remission 6 Died (8) A -35.6 22 1.88 20 1.21 3 Died (4) B -10.5 13 17 1.09 1.43 10 15 0.85 1.28 5 Died (12) A 25.0 65 2.72 . . 64 3.40 PET positron-emission tomography. * Elements of the German Society of Pediatric Oncology HIT-89. Numbers in brackets indicate how many days later the positron-emission tomography scan was performed. t Change in percentage between the highest tumor metabolic rate measured in the first and last positron-emission tomography study.

[1] 145w Brain tumors represent just under one-fifth of all childhood malignancies. Posterior fossa tumors are the most common primary malignant brain tumor entities in childhood. ' The addition of drug therapy to surgery and radiation therapy has improved the frequency and duration of recurrence-free survival in children with medulloblastoma and primitive neuroectodermal tumors (PNET), as published One of the dilemmas of current research in the treatment of medulloblastoma and PNET, however, is the difficulty of designing new clinical trials that are likely to increase the cure rate further. It would be of great interest to provide clinicians with a diagnostic tool to evaluate metabolic activity of childhood brain tumors and to assess the patient's early response to treatment at a time when modifications of the patient's management are possible. This could be one way to minimize long-term toxicity of treatment on intellectual, motor, and hormonal functions.','

[2] 102w A number of studies have evaluated the usefulness of positron emission tomography (PET) in brain tumor diagnostics.' It has been described that in adults, the level of glucose use by brain tumors correlates with the eventual clinical outcome of the patient." Hoffman et al." have published preliminary experiences showing that ['EF]2-deoxy-2-fluoro-D-glucose (FDG) uptake correlates with malignant tumor types in pediatric posterior fossa tumors. PET with FDG also enables clinicians to distinguish between recurrent tumors and regions of late tissue necrosis following radiation therapy in adult patients.'*,13 The use of labeled amino acids, such as ["Clmethionine, in adult patients has been shown to

[3] 272w Table 1. Summary of Clinical Data in Patients With Repeat Positron-Emission Tomography Studies After Operation (Group 1) Age Site of primary Tumor after Tumor location at 1st Treatment response: Patient (yr) Sex Histology tumor* resection PET scan' tumor mowth*t 4.5 M Medulloblastoma (classic type) 10.3 M Medulloblastoma 2.2 F Medulloblastoma 25.6 F Medulloblastoma (classic type) (classic type) (desmoplastic type) 0.5 F PNET 1.0 F PNET 13.3 M Medulloblastoma, partly medullomvoblastoma Vermis, left cerebellum Vermis, left pons Vermis to cervical Right cerebellum Left temporooccipital cortex, brain stem Edge ventricle I11 spine Vermis, edge ventricle IV Recurrent after subtotal resection and supratentorial metastases Recurrent after subtotal resection Recurrent after total gross removal Recurrent after total gross removal Recurrent after total gross removal Primary, stereotactic biopsy gross removal Recurrent after total Vermis Left cerebellum Right paneto-occipital cortex Left pons Brain stem Right cerebellum Brain stem Edge ventricle 111 Left cerebellum Vermis Smaller S m a 11 e r Disappeared Disappeared Disappeared Disappeared Disappeared Smaller Smaller Unchanged PET: positron-emission tomography; PNET primitive neuroectodermal tumor * Shown by computed tomography or magnetic resonance imaging scan. t At time of last positron-emission tomography scan. provide a better contrast between tumor and normal brain than PET with FDG.14 Recently published results have also suggested that the use of PET with ["Clmethionine in children with various brain tumors has a potential value in the differential diagnosis of tumor recurrence and radiation injury." Despite an increasing interest in the clinical application of PET in tumors of the adult patient as a diagnostic and prognostic tool, only few studies have been concerned with the usefulness of PET with FDG in childhood tumors.

[4] 39w The present pilot study was designed to evaluate whether PET measurements of tumor glucose metabolism have the potential to assess tumor response to treatment and to detect differences in the biologic activity of the most common childhood brain tumors.