Impact of proliferation index on outcome in childhood malignant gliomas: Results in a multi-institutional cohort

OBJECTIVE: Prognoses of pediatric high-grade gliomas are unpredictable, even when clinical and histological factors are taken into account. In preliminary studies with an institutional cohort of pediatric high-grade gliomas, we observed a strong association between outcome and proliferation index, as assessed by immunolabeling with the MIB-1 antibody. To determine whether this marker could provide prognostically useful information independent of tumor histology, we examined the prognostic usefulness of this marker in the multi-institutional cohort of Children's Cancer Group Study 945, the largest group of childhood high-grade gliomas analyzed to date. METHODS: The study group consisted of tumors within this cohort that were classified as high-grade gliomas on central review according to contemporary World Health Organization guidelines and that had sufficient histopathological material to permit proliferation index assessment. Paraffin-embedded sections were cut and processed, microwave antigen enhancement was used, and MIB-1 indices were calculated by percent labeling in approximately 2000 cells (5-10 high-power fields) in the areas with greatest labeling. To ensure that the review diagnostic classification and proliferation labeling index were assigned independently for each tumor, these analyses were performed by two different neuropathologists at separate institutions, and each was blinded to the results of the other. RESULTS: Ninety-eight tumors met eligibility criteria for this study. Among these high-grade gliomas, there was a strong association between MIB-1 labeling and patient outcome: 5-year progression-free survival was 33 +/- 7% in 43 patients whose tumors had MIB-1 indices of less than 18%, 22 +/- 8% in the 27 patients whose tumors had indices between 18 and 36%, and 11 +/- 6% in the 28 patients whose tumors had indices greater than 36% (P = 0.003). As anticipated, a strong association was also observed between histology and MIB-1 labeling index in these cases. Mean labeling indices were 19.4 +/- 2.66 for tumors classified as anaplastic astrocytoma versus 32.1 +/- 3.08 for those classified as glioblastoma multiforme (P = 0.0024). Notwithstanding this correlation, a significant association was noted between labeling index and progression-free survival, even after the analysis had been stratified by histology (P = 0.001). Although histology had an independent association with outcome, the prognostic value of MIB-1 labeling transcended histological subgrouping and was apparent both in tumors classified as anaplastic astrocytoma (P = 0.02) and in those classified as glioblastoma multiforme (P = 0.046). Multivariate regression modeling confirmed the strong independent association between MIB-1 labeling index and outcome. As a group, tumors with labeling indices higher than 36% had an almost uniformly poor outcome, regardless of histology. CONCLUSION: MIB-1 labeling index and histological categorization are each prognostically relevant in childhood high-grade gliomas. MIB-1 labeling index can help to refine the accuracy of histologically based prognostic assessments.