Wnt/Wg pathway activation in medulloblastoma and disease risk stratification

Abstract

Despite improvements in the diagnosis and therapy of medulloblastoma (MB), the most common malignant paediatric brain tumour, survival rates are still only 40-60% for high-risk patients, and surviving patients offer suffer life-long therapy-associated late effects. Improving the current clinical disease risk classification may provide opportunities to tailor therapy to disease-risk more accurately, aiming to improve survival whilst minimizing late effects. The overall aim of this study was to identify and validate clinicopathological and molecular markers of prognostic significance in a large cohort of cases derived from the PNET3 clinical trial, which could be used in improved disease stratification schemes.

Initial investigations focussed on defects of chromosome 17, the most common chromosomal aberrations identified in MB, which have been associated with adverse outcome in some previous studies. During this initial study, methods for testing for regions of genetic loss in DNA samples extracted from routinely collected and stored pathology material were developed and validated. Using these methods, 17p loss was assessed in 284 cases overall, but did not show an association with poor prognosis. 17p loss was rare in patients <3 years at diagnosis (1/21 cases) compared to children aged >3 years (21/59, p=0.009), supporting the notion that infants with MB may represent a biologically distinct disease subgroup. During the course of this study, evidence of a microsatellite instability phenotype was also found in one case (1/29; 3.8%), suggesting a role for this phenomenon in a small subset of medulloblastomas.

Previous studies have identified MBs with activation of the Wnt/Wg signalling pathway to be associated with a favourable outcome. Three independent molecular markers of Wnt/Wg activation in medulloblastoma (β-catenin nuclear localization, CTNNB1 mutation and chromosome 6 loss) have recently been identified, and were therefore assessed in the PNET3 cohort to examine their inter-relationships and relative prognostic utilities. β-catenin nuclear localization was detected in 34/207 (16.4%) cases, CTNNB1 mutation in 20/197 (10.1%), and chromosome 6 LOH in 20/190 (10.5%); all three markers were highly associated with each other (p=0.001). Univariate survival analysis showed significant associations between clinicopathological (gender, pathology, M stage and extent of tumour resection) and molecular variables (β-catenin nuclear localization and CTNNB1 mutation) with survival (all 'p' values <0.03) in the PNET3 cohort. Multivariate survival analysis and Cox proportional regression hazard analysis identified M stage, large cell / anaplastic pathology, gender and β-catenin nuclear staining to be significantly associated with prognosis (p<0.04), with hazard ratios of 1.924, 3.338, 0.57 and 0.215, respectively. Finally, combined models for disease risk prediction were constructed using the prognostically significant clinicopathological and molecular variables identified, including models to precisely classify the overall disease hazard for individual cases based on their clinical, pathological and molecular profile. In summary, Wnt/Wg pathway activation was associated with favourable outcome and can be most accurately identified using β-catenin nuclear staining. Combined models which incorporate Wnt/Wg status alongside other prognostically significant clinical and pathological indices validated in this study provide schemes for improved disease risk prediction in medulloblastoma, which should now be tested prospectively in large clinical trials.