Which clinical and biological tumor markers proved predictive in the prospective multicenter trial HIT'91--implications for investigating childhood medulloblastoma

BACKGROUND: Most recent studies analyzing candidate biological prognostic factors in childhood medulloblastoma (MB) are limited by small patient numbers due to dependence on fresh-frozen tumor material. By contrast, large archives of formalin-fixed, paraffin-embedded MB samples exist from homogeneously treated patients. PATIENTS AND METHODS: We have optimized RNA and DNA isolation from formalin-fixed paraffin-embedded MB samples. We then analyzed archived tumor samples from well-documented patients treated within the prospective randomized multicenter trial HIT'91 for DNA amplification of c-myc and N-myc, and mRNA expression of c-myc and trkC. RESULTS: TrkC and c-myc mRNA expression were identified as independent prognostic factors by multivariate analysis. Three risk groups were identified: 1) Favorable risk group: All 8 patients (2 metastatic) with elevated trkC and reduced c-myc mRNA expression (compared to levels of human cerebellum) remained relapse-free (7-year EFS 100%). 2) Poor risk group: 10 of 15 patients with metastatic disease and high c-myc and low trkC mRNA expression relapsed (7-year EFS 33%). 3) Intermediate risk group: The 7-year EFS of the remaining 78 patients was 65%. CONCLUSIONS: While the collection of fresh-frozen tumor samples is remaining a major challenge in large clinical trials, routinely processed paraffin-embedded tissue samples can be used to quantitate biological prognostic factors on the DNA and RNA level. Upon prospective validation of cut-off levels, this may lead to better risk-based stratification systems for children with medulloblastoma.