Prospective validation of a molecular prognostication panel for clival chordoma

OBJECTIVE There are currently no reliable means to predict the wide variability in behavior of clival chordoma so as to guide clinical decision-making and patient education. Furthermore, there is no method of predicting a tumor's response to radiation therapy. METHODS A molecular prognostication panel, consisting of fluorescence in situ hybridization (FISH) of the chromosomal loci 1p36 and 9p21, as well as immunohistochemistry for Ki-67, was prospectively evaluated in 105 clival chordoma samples from November 2007 to April 2016. The results were correlated with overall progression-free survival after surgery (PFSS), as well as progression-free survival after radiotherapy (PFSR). RESULTS Although Ki-67 and the percentages of tumor cells with 1q25 hyperploidy, 1p36 deletions, and homozygous 9p21 deletions were all found to be predictive of PFSS and PFSR in univariate analyses, only 1p36 deletions and homozygous 9p21 deletions were shown to be independently predictive in a multivariate analysis. Using a prognostication calculator formulated by a separate multivariate Cox model, two 1p36 deletion strata (0%-15% and > 15% deleted tumor cells) and three 9p21 homozygous deletion strata (0%-3%, 4%-24%, and >= 25% deleted tumor cells) accounted for a range of cumulative hazard ratios of 1 to 56.1 for PFSS and 1 to 75.6 for PFSR. CONCLUSIONS Homozygous 9p21 deletions and 1p36 deletions are independent prognostic factors in clival chordoma and can account for a wide spectrum of overall PFSS and PFSR. This panel can be used to guide management after resection of clival chordomas.