Biomarker-driven stratification of disease-risk in non-metastatic medulloblastoma: Results from the multi-center HIT-SIOP-PNET4 clinical trial

Abstract

Purpose: To improve stratification of risk-adapted treatment for non-metastatic (M0), standard-risk medulloblastoma patients by prospective evaluation of biomarkers of reported biological or prognostic significance, alongside clinico-pathological variables, within the multi-center HIT-SIOP-PNET4 trial.

Methods: Formalin-fixed paraffin-embedded tumor tissues were collected from 338 M0 patients (>4.0 years at diagnosis) for pathology review and assessment of the WNT subgroup (MBWNT) and genomic copy-number defects (chromosome 17, MYC/MYCN, 9q22 (PTCH1) and DNA ploidy). Clinical characteristics were reviewed centrally.

Results: The favorable prognosis of MBWNT was confirmed, however better outcomes were observed for non-MBWNT tumors in this clinical risk-defined cohort compared to previous disease-wide clinical trials. Chromosome 17p/q defects were heterogeneous when assessed at the cellular copy-number level, and predicted poor prognosis when they occurred against a diploid (ch17(im)/diploid(cen)), but not polyploid, genetic background. These factors, together with post-surgical tumor residuum (R+) and radiotherapy delay, were supported as independent prognostic markers in multivariate testing. Notably, MYC and MYCN amplification were not associated with adverse outcome. In cross-validated survival models derived for the clinical standard-risk (M0/R0) disease group, (ch17(im)/diploid(cen); 14% of patients) predicted high disease-risk, while the outcomes of patients without (ch17(im)/diploid(cen)) did not differ significantly from MBWNT, allowing re-classification of 86% as favorable-risk.

Conclusions: Biomarkers, established previously in disease-wide studies, behave differently in clinically-defined standard-risk disease. Distinct biomarkers are required to assess disease-risk in this group, and define improved risk-stratification models. Routine testing for specific patterns of chromosome 17 imbalance at the cellular level, and MBWNT, provides a strong basis for incorporation into future trials.

Keywords: biomarker; clinical trial; medulloblastoma; stratification.

Figure 1. MB_WNT subgroup tumors: clinical, pathological and molecular correlates

A. Examples of tumors scored negative and positive for β-catenin nuclear accumulation, with CTNNB1 status also indicated. B. Distribution of clinical, histopathological and molecular markers between the MB_WNT (nuclear β-catenin accumulation) and non-MB_WNT medulloblastoma subgroups. C. Age distributions of MB_WNT and non-MB_WNT patients. D. Kaplan-Meier plots and associated ‘p’ values (log-rank test) shown for MB_WNT patients with age at diagnosis younger than 16 years, older than 16 years versus non-MB_WNT cases.

Figure 2. Prognostic significance of molecular disease features in the HIT-SIOP-PNET4 cohort

Kaplan-Meier plots and associated ‘p’ values (log-rank test) are shown for each feature. Chromosome 17 losses and gains observed against a diploid centromeric reference score (diploid(cen)) are represented.

Figure 3. Chromosome 17 defects in HIT-SIOP-PNET4 cohort tumors

Patterns (A, C) and prognostic significance (B.; ‘p’, log-rank tests) of chromosome 17 defects detected by iFISH. C. iFISH analysis showing (i) 17p loss (single green signals) and (ii) 17q gain (three green signals) against a diploid centromeric background (two red signals). Nuclei are counterstained blue. D. Relationship of ch17(im)/diploid(cen) defects to clinico-pathological and molecular disease features assessed (‘p’, Fisher's exact or χ² tests; corrected and uncorrected values are shown). Abbreviations: ch, chromosome; im, imbalance (p-gain and/or q-loss); diploid(cen), diploid centromeric signal; polyploidy(cen), polyploidy centromeric signal.

Figure 4. Biomarker-driven risk-stratification models for standard-risk (M0/R0) medulloblastoma based on patients from the HIT-SIOP-PNET4 cohort with data available for all parameters (n = 118)

A. Established disease-wide survival model for non-infant medulloblastoma [1, 10, 14] (LCA pathology and/or MYC/MYCN amplified, high-risk; MB_WNT and no high-risk features, favorable-risk; others, intermediate-risk). B. Empirically-derived survival model for non-infant, standard-risk medulloblastoma. C. Illustrative survival model for non-infant, standard-risk medulloblastoma, incorporating the distinction of MB_WNT patients into the empirically-derived model. Kaplan-Meier plots and associated ‘p’ values (log-rank tests) show EFS. M0, non-metastatic; R0, no significant post-surgical tumor residuum; ch17(im), ch17(im)/diploid(cen) tumors. D. Time-dependent receiver operator characteristic (ROC) curves showing predictive performance of the three models for survival at five-years, determined as the area-under-curve (AUC). TP, true positive; FP, false positive. Chromosome 9 defects were not assessed in survival modelling due to missing data points.