Drs. Jubran and Finlay have written a timely review that articulates the trials and tribulations of treating central nervous system (CNS) germ cell tumors. Their review comes on the heels of the Children's Oncology Group (COG) nongerminomatous germ cell tumor study (ACNS0122) that opened January 26, 2004, and in anticipation of the COG germinoma study (ACNS0232) that may open in 2005. Incidence
CNS germ cell tumors are rare among brain tumors affecting adults and children. Based on data from the Central Brain Tumor Registry of the United States,[1] fewer than 93 are expected to be diagnosed annually in US males under age 19, and the number of tumors diagnosed in females is expected to be less than 38. These numbers demonstrate the male preponderance of this disease, and given the uncertainty in the classification of these tumors, the numbers are rough estimates. Progress in the treatment CNS germ cell tumors has been slowed by the number of patients available for protocol enrollment, the diversity of tumor histiotypes, the multitude of clinical presentations, the range of disease extent at diagnosis, the presence or absence of tumor markers, and the variable radiographic characteristics. As noted by Drs. Jubran and Finlay, the approach to the treatment of these tumors can be separated early into the germinomatous and nongerminomatous germ cell tumor groups with little overlap. Germinoma, analogous to seminoma in the extra-CNS setting, is highly sensitive to all forms of therapy, making decisions regarding management difficult with a major emphasis on side effects of treatment instead of disease control. Nongerminomatous germ cell tumors, in contradistinction to their extra-CNS analogs, are less sensitive to all forms of therapy, which serves to focus the goals of investigators and makes disease control a priority over treatmentrelated side effects. Staging
The importance of staging cannot be overemphasized in the management of these patients. Inadequate staging has been the identifiable downfall for a number of studies worldwide and has confounded the interpretation of results. Poor staging affects classification for outcomes analysis, response evaluation, radiation dose and volume prescriptions, and, clearly, the cure of these patients. High-quality spinal magnetic resonance imaging is required as well as adequate cerebrospinal fluid (CSF) sampling. The negative predictive value of a single lumbar CSF cytologic analysis remains unclear and should be considered an important secondary objective in any major study. Serum and CSF alpha-fetoprotein and beta-HCG are important for diagnostic and staging purposes, including response evaluation and decisionmaking as to the use of surgery and radiation therapy. However, each institution and cooperative group has its own criteria for cutoff values and risk classification; ratios of serum to CSF values and the normal values in the CSF have not been established. Standardization in the use of serum and CSF markers-as well as the assays used for their measurement-is required. In addition, markers such as carcinoembryonic antigen, placental alkaline phosphatase, lactate dehydrogenase, and the soluble form of c-kit, a transmembrane tyrosine kinase receptor, should be investigated.[2] The diagnosis of germinoma should require a biopsy since the objective of most treatment regimens is to minimize therapy. Although their radiographic appearance and location may be characteristic, because of the absence of markers and the differences in management between germinoma and other tumor types, a tissue diagnosis should be established. As pointed out by Drs. Jubran and Finlay, histopathology in the setting of elevated markers is not necessary. However, when the markers are negative, when beta-HCG is low but elevated, or when there is any question concerning the diagnosis based on radiographic characteristics or clinical presentation, we favor biopsy if it can be performed with a low risk of side effects. Although most biopsies are extremely small and provide limited detail regarding the heterogeneity of these tumors, tissue can be invaluable when the response to therapy is poor or mixed and when second surgery is performed. Chemotherapy
Based on the curability of extra- CNS germ cell tumors with multiagent chemotherapy, it was logical for investigators to consider chemotherapy alone as a treatment option. This option was pursued for several years in light of concerns about the effects of radiation therapy. Researchers have explored two chemotherapeutic approaches to CNS germ cell tumors: chemotherapy alone and chemotherapy as a component of a combined- modality strategy involving the use of reduced dose and volume irradiation. Investigators have attempted to treat patients with chemotherapy alone in two successive trials.[3,4] These trials differed in the types of agents used as well as their intensity and sequencing. In the first trial, eventfree survival was short, the progression rate was high, and there were a noticeable number of toxic deaths. In the second trial, outcomes were similarly poor and toxicity remained high. The key finding from these trials is that a subset of patients may be effectively treated without irradiation and that salvage with radiation therapy- at least for patients with germinoma- may be feasible. We need to determine which patients might be candidates for avoiding irradiation and consider this treatment approach only for the youngest patients. Toxic deaths associated with the use of chemotherapy alone have made a bad impression; however, it has been acknowledged that such studies, which were internationally based, were carried out at centers that might not be equipped to handle this type of patient and treatment intensity. Nongerminomatous Germ Cell Tumor Study
There are too many combinedmodality approaches for CNS germ cell tumors to list, and hopefully the current and proposed studies through COG comprise the best of all prior experiences. The aim of the current nongerminomatous germ cell tumor study being conducted by COG is to assess the ability of neoadjuvant chemotherapy, with or without preirradiation surgery, to reduce or eliminate measurable disease prior to radiation therapy. Patients enrolled in this trial receive alternating combinations of carboplatin/ etoposide and ifosfamide/ etoposide every 3 weeks for 18 weeks (induction) followed by response assessment. Patients with progressive disease are taken off protocol; those with a complete response proceed to craniospinal irradiation that includes neuraxis irradiation to 36 Gy and a focal boost to 54 Gy using a 1-cm clinical target volume margin based on assessment of pre- and postinduction neuroimaging. Those with less than a complete response have the option of surgical resection. Those with more than 65% reduction in disease volume (partial response) after induction chemotherapy or induction chemotherapy and preirradiation surgery will proceed to craniospinal irradiation provided that their markers have normalized. Those with stable disease or less than a partial response and abnormal markers receive consolidative chemotherapy using thiotepa/ etoposide and peripheral stem cell rescue prior to radiation therapy. From the participant's point of view, this protocol is relatively straightforward and seeks to establish a new baseline for these patients using conventional chemotherapy up front, modern neurosurgical approaches, and relatively standard radiation therapy. The combined complete and partial response rates are expected to be above 70% for this study group and will hopefully translate into improved outcomes. CNS Germinoma Study
The upcoming COG study for CNS germinoma is ambitious, as it proposes to randomize patients between "standard" radiotherapy and combined- modality therapy consisting of two cycles of carboplatin and etoposide (induction) followed by radiotherapy (in complete responders) or further chemotherapy and radiotherapy based on response assessment (mainly in those with minimal residual disease, partial responses, or stable disease). As proposed, there are at least eight possible treatment groups combining the different possible stages of disease, responses, and randomization arms. Standard radiotherapy for patients with localized disease is ventricular irradiation to 24 Gy followed by focal irradiation with 21 Gy, for a total primary site dose of 45 Gy. Standard radiotherapy for patients with disseminated disease is craniospinal irradiation to 24 Gy followed by focal irradiation with 21 Gy, for a total primary site dose of 45 Gy. As detailed by Drs. Jubran and Finlay, the lack of consensus with regard to treatment volumes, the noticeable risk of failure outside the irradiated volume in many series, and questionable adequacy of staging procedures has made decisions about radiotherapy treatment guidelines controversial.[5] Patients treated in the combinedmodality arm of the COG CNS germinoma study will receive responseand risk-adapted radiation therapy, including focal irradiation to 30 Gy for patients with localized disease in complete response after induction, and focal irradiation to 40.5 Gy for patients with a partial response after induction and complete response or minimal residual disease after two additional cycles of cisplatin and cyclophosphamide. Variations to the proposed regimen are included for those with neuraxis dissemination at diagnosis. They will receive combined-modality treatment with craniospinal irradiation to 24 Gy, unless they have a complete response to induction, in which case 21 Gy will be administered. Other variations in the protocol regimen are included for patients who have progressive disease or who do not respond to chemotherapy after induction. Unanswered Questions
The COG study for CNS germinoma hopes to answer a number of important questions about the ability of chemotherapy to treat microscopic residual disease and to reduce the dose of irradiation, the prediction of outcome based on response to chemotherapy, the adequacy of ventricular irradiation, and the possibility of differentiating neurocognitive outcomes for patients who receive radiotherapy alone or via a combined-modality approach. A number of these questions have been addressed previously but, for the most part, not in the context of a large randomized trial. In addition, the toxicity of radiation therapy or combinedmodality therapy has never been prospectively assessed in such a manner as to guide treatment decisions. With increasing levels of communication among investigators on all continents, and with the future convergence of opinions regarding diagnosis, staging, surgery, chemotherapy, and radiation therapy, a future international protocol with broad participation may be feasible.