Cancer Research for Children
Project location: ITALY
Project start date: January 2005 - Project end date: This project covers various years
Project number: 2005-13
Beneficiary: AIRC - Associazione Italiana per la Ricerca sul Cancro
[2009-01] Timeline of the present activity report: 2008-2010
Task 1. Diagnosis, treatment and follow-up, of Italian patients on a European
background. One-hundred and nine newly diagnosed patients have been treated so far. The total number of cases corresponds to 88% of those registered in the data-base of AIEOP (Associazione Italiana di Ematologia ed Oncologia Pediatrica) with a supposed diagnosis of ependymoma and this percentage has been increasing along the years. Among the 109 patients accrued so far, 27 had less than 3 years at time of diagnosis: 25 received post-operative adjuvant treatment according to the known guidelines while 2 children have been only followed-up because of grade 2 histology and complete tumor resection. It is also worthwhile to outline that the model of front-line review in ependymoma has been extended to all the pediatric neuro-oncology cases in Italy. Of the 109 children treated so far, 64 were male and 45 female with a median age of 5 years; tumor was located in posterior fossa in 69 and in supratentorial areas in 40. Central histological review assigned grade 3 to 59 cases and grade 2 to the other 50; tumor was completely resected in 72 cases and had a residual as evaluated by post-operative MRI in 37 patient, one of those with metastases. According to protocol guidelines, a second-look surgery was to be attempted in all cases presenting with tumor residue after a first surgery. In 26 children the residue was potentially eligible for surgical re-treatment. As final result, 12 of these 26 children were tumor-free after more than one surgical excision while the others had a cytoreduction. Two patients reported facial and acoustic nerve hypofunction omolateral to the lesion after second-look surgery. Radiotherapy boost was applied after fractionated treatment to 14 children whose tumor residue was still measurable after one or more surgical acts before adjuvant treatment: it was not followed by severe side-effects. Considering all the patients, 28 of them relapsed so far at a median of 19 months (range 5-72) after diagnosis, 16 have died. The median follow-up of the series is 34 months and the EFS/OS at three years are 69.2 5.5 % and 83.4 4.4% respectively. This treatment protocol is feasible, the accrual is satisfying and toxicity affordable. The two more severe prognostic categories that are patients with tumor residual and with anaplastic histology were treated differently in this protocol compared to the first one: all received radiation therapy at a total dose of 59.4 Gy delivered at daily fraction of 1.8 Gy, all received chemotherapy according to VEC schedule (previous to second-look surgery and radiation or after radiation if rendered tumor-free by surgery), and patients with post-surgical residual had also the addition of radiation boost. At present time there are no statistical differences according to residual disease and grade, thus showing the prognostic amelioration in these subgroups. Also children under three years of age at diagnosis are having the same results while their outcome was previously discouraging.
Among the points in discussion in our treatment strategy, these efforts have as final goal the generation of a new European treatment protocol that will be probably opened next January:
• For all other children, RT (59,4Gy, one fraction daily) will be the standard.
• The randomised question will be the role of chemotherapy after radiation for patients without tumor: VEC was defined as an adequate back bone, to which Cisplatin and Vincristine have been added.
• For those with residue, window therapies are warranted. Then radiation therapy at a total dose of 59.4 Gy and a boost (8 Gy in 2 fractions) will be delivered if a second surgery has been excluded.
Task 2. Biological studies ancillary to the clinical tasks.
Validation of reported and novel candidate prognostic markers.
We constructed, from 41/60 ependymoma samples of prospectively treated patients, a tissue microarray (TMA). Immunohistochemistry evaluated Nucleolin and EGFR, FISH evaluated 1q gain. Moreover, for 29 tumors with frozen fresh material, the mRNA expression of EGFR, prominin1/CD133 cancer stem cell marker and telomerase catalytic subunit (HTERT) was evaluated by quantitative RT-PCR. Among clinical variables, at multivariable analysis by Cox model, only grading III vs II affected both EFS (p=0.05) and OS (p=0.07). Multivariable analyses of each biological marker separately accounting for penalty-weighted estimates were performed with adjustment for clinical variables. Relatively high hazard ratios (HRs) for EFS were achieved at the protein level by Nucleolin (HR high vs low: 1.88; p=0.110) and EGFR (HR low vs strong: 1.76; p=0.140), while at the mRNA level by EGFR (HR 32.7 vs 5.1: 1.76; p=0.396) and HTERT (HR 33.5 vs 1.2: 2.04; p=0.337); the 5-year EFS was 46% vs 20% (P=0.004) in the subgroup with low (<50%) versus high Nucleolin, respectively. The strongest association with OS emerged for HTERT mRNA expression (HR: 9.9; p=0.011), while EGFR (HR: 3.12; p=0.444) was not significant. The 5-year OS was 84% vs 48% (p = 0.005) in the subgroup with hTERT<6 median value versus 6, respectively. In this clinical and biological evaluation multivariable analyses show that Nucleolin was the strongest biological marker for EFS while HTERT emerged as the strongest predictor for OS.
Biological consortium through Europe has already provided the possibility of sharing tumor samples and performing meta-analyses. These results, as well those performed on Italian microarray, will be presented at the forthcoming ISPNO.
In the European context, potential prognostic markers (metallothioneins 1-2 and 3, YAP1, nucleolin/telomerase, ASPM, EGFR, KI-67, tenascin-C) were initially tested by immunohistochemistry in tumours from patients treated in three countries (United Kingdom n=105, Italy n=62, France n=93). Young children were treated with chemotherapy first whileolder children received irradiation plus chemotherapy in case of incomplete resection. Based on national results, the most promising markers were then tested on TMA slides for the entire three cohorts. Prognostic value of markers for OS and PFS were assessed through a Cox model stratified on country, testing simultaneously localization, grade, extent of surgery and treatment. Median age of the population included was 34 months. Median follow-up was 7.8 years. Factors associated with a higher risk of death were a strong expression of tenascin-C (HR=2.13, 95% CI=[1.36-3.35], p=0.001), incomplete surgery (HR=1.79 [1.17-2.75], p=0.007) and grade (HR=1.61 [0.99-2.62], p=0.05). Factors associated with a higher risk of recurrence were incomplete surgery (HR=1.53 [1.09-2.15], p=0.01) and treatment other than irradiation (p=0.004). A trend in association between strong expression of tenascin-C and risk of recurrence was observed but was not significant (adjusted HR=1.39 [0.97-1.98], p=0.07).
Strong expression of tenascin-C is independently associated with a higher risk of death in paediatric ependymoma, adjusted on extent of surgery and grade. Only the treatment approach was prognostic of recurrence-free survival. On this way we are therefore discovering potential prognostic markers that need to be valuated in prospective series such as the future SIOP European protocol that will put together our clinical and biological efforts.
In order to investigate additional novel potential molecular markers for prognosis and therapeutic purposes, we investigated the microRNA and methylation profiling of primary and recurrent ependymomas (n= 59), low-grade gliomas (n= 30) and embrional tumors (n= 12).
Unsupervised hierarchical clustering of gene expression data robustly distinguishes the different tumor types and class comparison analyses define ependymoma-specific genes signatures (p < 0.001). A miRNA signature of 10 genes suggestively discriminates primary versus relapsing ependymoma (p = 0.07 for global test accounting for multiple testing), mir- 10a and 10b are frequently de-methylated and mir-10a is highly expressed at relapse. No significant associations with overall survival were identified. Methylation profiling identified RASSF1A epigenetic inactivation in infratentorial tumors, further supporting the observation that NF2, YAP1 and RASSF1A components of the Salvador-Warts-Hyppo pathway are differentially altered in all ependymoma subsets. We also identified IGF2 loss-of-imprinting and concurrent high expression of IGF1R, suggesting the existence of an aberrant
autocrine/paracrine insulin signalling.
Assessing the role of conventional and novel targeted therapies.
Matched frozen tissue and short-term cell cultures from four ependymoma patients were extensively analyzed for the expression of genes involved in response to conventional treatments (ABC transporters, chemo/radio resistance genes) as well as in growth factor mediated pathways. Ependymoma cell cultures mirror the expression profiles identified in frozen and paraffin-embedded tumor tissue, including overexpression of components of EGFR, YAP1 and NOTCH pathways. In parallel, short-term cell cultures were used for pharmacologic studies using both conventional (cisplatin, topotecan, etoposide, vincristin, BCNU) and targeted agents. The response to drug agents revealed that ependymoma cells are generally resistant to treatments, as demonstrated by high IC50 values compared to unrelated cell lines used as control, and gene expression analyses revealed elevated expression of chemoresistance-related genes. In addition, analysis of gene expression of cisplatin-treated cells revealed a partially impaired apoptotic signalling, in particular lacking activation of caspases genes, despite the wild type status of TP53 gene is retained in the ependymoma cell lines analyzed. Interestingly, the expression of a set of genes related to chemo/radio resistance (ALKBH2, DEPC, ABCC1, GSTP1, ABCB1, TOP2A, PROM1) was analyzed in a total of 65 ependymoma frozen samples, including primary and relapsing cases.
TOP2A gene was upregulated at the mRNA level but lack of TOP2A protein expression in the ependymoma cell cultures correlates with the observed limited efficacy of topotecan treatment. We investigated the correlation with the clinical parameters (grade, age, status) and, notably, only young age showed a highly significant correlation with gene expression (ALKBH2 and ABCC1 genes, p-value <0.01). Instead, relapsing tumors showed only moderately increased levels of expression of genes related to DNA repair mechanisms, further suggesting that tumor relapse is not prominently due to onset of treatment resistance, but instead to intrinsic treatment insensitivity. Finally, despite elevated expression of multiple receptor tyrosine kinases as well as the observed constitutive AKT activation and cytotoxic response to PI3K inhibitor LY294002 in ependymoma cells, the antiproliferative effect of several receptor-kinase inhibitors individually evaluated was limited.
Production of in vitro and in vivo mouse models of ependymoma.
No in vitro and in vivo preclinical models of ependymoma are available world-wide except for short-term cell cultures. In light of the important role of HTERT expression for ependymoma patient survival, as we and others have shown, and given the importance of HTERT gene for cancer cell stable proliferation in vitro, we analyzed HTERT expression in our short term cell cultures and identified decreasing levels upon serial passaging. Similar to other cancer types (Dalerba, Cancer Res 2005, 65:2321-9), we speculate that transduction of these cells with a retroviral vector encoding HTERT is sufficient to reconstitute telomerase activity and allow immortalization and therefore have established HTERT-overexpressing ependymoma cultures, which will be tested as long-term, robust pre-clinical models.
Task 3 Neuropsychological evaluation/ Observation and psychological intervention/
We have investigated cognitive and psychological problems in children treated for ependymoma, the evolution of these disorders over time and the role of radiotherapy, age at radiotherapy and tumor site, in their onset and persistence.
Twenty-three patients received a complete evaluation; some of them underwent follow-up reevaluation.
The clinical data collected included sex, age (at diagnosis, assessment and tumor treatment), histological diagnosis, site (supratentorial vs. subtentorial), the presence of hydrocephalus, neurological examination, tumor treatment and radiotherapy protocol. All the patients received an age-appropriate cognitive and psychological evaluation.
The mean cognitive level was within the norm, with lower scores on the PIQ. The psychological assessment revealed Internalizing problems and impaired independence. No differences in outcome were found between the radiation adopted in the first and in the second protocol. The older children had a lower Intellective Quotient (IQ) than the younger children, both at the initial evaluation and at follow-ups. Initially, the supratentorial group appeared to be less impaired than the subtentorial group but then exhibited a progressive decline in the IQ. In the subtentorial group, the children with an IQ within the normal range remained stable at the follow-up, while the children with below-norm scores at the initial evaluation showed deterioration over time. Concluding observations are that site and pre-post operative damage seem to affect the cognitive outcome to a greater extent than age at radiotherapy.