Ian F. Pollack, MD

1) Molecular Markers of Prognosis in Gliomas

Malignant astrocytomas are among the most common and deadly brain tumors of childhood. Most affected children die within several years of diagnosis, despite current treatments; however, 20 to 30% respond favorably to therapy and are cured. The basis for these diverse outcomes has been enigmatic, even taking into account clinical and histological factors. In preliminary studies with an institutional cohort of pediatric gliomas, we observed that molecular markers could supplement histological data to refine prognostic assessments. Based on these findings, we began a more extensive study of the cohort of Children’s Cancer Group study CCG-945, the largest group of pediatric high-grade gliomas accrued to date, and subsequently the Children’s Oncology Group ACNS0126 study. The large size of these cohorts and the consistent treatment approaches used, coupled with the availability of central neuropathology review and comprehensive clinical data, provided a unique opportunity to address issues of molecular etiology and prognostic factors. Our studies demonstrated a striking association between outcome and several molecular features, including MGMT expression status, independent of clinical or histological factors; identified significant differences between molecular features of childhood and adult gliomas; and generated a sizeable resource of tumor tissue for further analyses. Our ongoing studies will use this resource as well as newly acquired tumor specimens and paired normal tissue, derived from children treated on two new Children’s Oncology Group high-grade glioma studies, to define the genetic alterations that characterize pediatric malignant gliomas, as a basis for prioritizing novel targets for prognostic and therapeutic stratification. We hypothesize that categorization of these tumors by their genomic alterations and drug resistance phenotype will improve accuracy of diagnostic and prognostic assessments, and provide insights into novel therapeutic targets. To test these hypotheses, we have initiated studies with the following aims: 1) Assess the frequency, composition, and prognostic relevance of alterations in Akt and MAPK pathway activation in pediatric malignant gliomas; 2) Determine whether O6-methylguanine-DNA methyltransferase (MGMT) overexpression and promoter methylation are associated with progression-free survival, independent of alkylator therapy; 3) Define, on a genome-wide basis, the spectrum and prognostic relevance of genomic alterations in pediatric malignant gliomas, using high-density SNP microarray-based profiling. Relevant markers will be evaluated in the context of conventional prognostic factors, such as histology, to determine their utility for biologically classifying childhood malignant gliomas. Taken together, these studies will incorporate a unique resource of childhood malignant brain tumor samples to provide new insights into the molecular categorization of pediatric high-grade gliomas. This work will establish a foundation for risk-adapted stratification and treatment planning, and the design of future therapeutic strategies for children with these tumors.

2) Molecularly Targeted Therapies for Malignant Gliomas

The limited response of malignant gliomas to conventional therapy reflects resistance to undergoing apoptosis in response to DNA damage or mitogen depletion, resulting from tumor suppressor gene mutations and aberrant activation of growth factor signaling. However, our previous studies indicated that despite the limitation in apoptotic triggering, effector pathways of apoptosis may remain intact and can be activated by inhibiting growth factor-mediated signaling or stimulating death receptor pathways. These studies also demonstrated that although a subset of gliomas were responsive to modulation of individual signaling pathways, many showed incomplete growth inhibition, reflecting activation of parallel pathways or intrinsic resistance mechanisms. This led us to examine the efficacy of combinatorial strategies for signaling inhibition, using agents targeting distinct pathways. Our initial studies suggested the potential for intriguing, synergistic interactions between signaling modulatory approaches, such as inhibition of PKC and Raf or JAK/STAT, and activation of apoptotic signaling by TRAIL, and with conventional therapies. Based on our findings, we hypothesized that therapeutic approaches that block rationally selected combinations of growth signaling pathways or that enhance apoptosis signaling will provide a novel strategy for inducing glioma cytotoxicity. To test this hypothesis, we are examining the effects on glioma growth and viability of inhibiting combinations of parallel pathways that transmit proliferative signals from aberrantly activated upstream receptors. These studies incorporate a panel of cell lines with defined genetic alterations to assess whether genotypic features influence efficacy, and establish biological surrogates of response. Second, we are determining whether signaling mediators that promote caspase expression can enhance apoptosis induced by stimulation of death receptor pathways by TRAIL, and evaluating biological factors that predict efficacy. Third, we are examining whether signaling modulation can enhance efficacy of radiotherapy and conventional chemotherapy in all, or a genotypically defined subset of, gliomas. These studies will provide a foundation for the translation of signal transduction inhibition and death receptor activation as therapeutic approaches for gliomas, and indicate ways in which these strategies can be used to enhance efficacy of other therapies.

3) Vaccine Therapy for Pediatric Gliomas

Diffuse brainstem gliomas, other malignant astrocytomas, and multiply recurrent low-grade gliomas carry a poor prognosis, and new therapies are needed. Having gained experience with immunotherapy for adult gliomas, we extended these insights to childhood gliomas, based on our observations regarding their profiles of glioma-associated antigen (GAA) expression. We initiated a pilot trial of subcutaneous vaccinations with peptides for GAA epitopes emulsified in Montanide-ISA-51 given every 3 weeks for 8 courses along with intramuscular injections of poly-ICLC in HLA-A2+ children with newly diagnosed brainstem gliomas (BSG), high-grade gliomas (HGG), or recurrent gliomas. GAAs were EphA2, IL13Rα2, and survivin. Primary endpoints were safety and T cell responses against vaccine-targeted GAAs, assessed by ELISPOT analysis. Treatment response was evaluated clinically and by MR imaging. To date, 28 children have been enrolled and assessed for response, 16 with newly diagnosed BSG, 5 with newly diagnosed HGG, 4 with recurrent low-grade gliomas and 3 with recurrent HGGs. No dose-limiting non-CNS toxicity has been encountered. One child with a BSG had transient tumor enlargement in association with acute neurological deterioration 4 months after beginning vaccination that later regressed and culminated in a sustained partial response (PR), consistent with pseudoprogression. Two other children with BSG had symptomatic pseudoprogression, with transient neurological deterioration and tumor enlargement followed by stabilization on decreasing steroid doses. Among 24 patients evaluable for response, 3 had rapidly progressive disease, 16 had stable disease for > 2 cycles, 3 had PRs, 1 had an MR, and 1 had prolonged disease-free status after surgery. ELISPOT analysis, completed in 13 children, showed GAA responses in 11, most commonly to IL13Rα2. We conclude that peptide vaccination in children with gliomas is generally well tolerated, although distinguishing pseudoprogression from true progression can be challenging. Immunological and clinical responses have been obtained. More extensive analyses of efficacy in a multi-institutional context are warranted.