Prospective Validation of Molecular Markers Predicting Overall Progression Free Survival, and Progression Free Survival after Radiation in Clival Chordomas

Background: Clival chordomas are rare and locally aggressive tumors with high recurrence rates despite surgery and radiation. Molecular prognostication for these malignancies remains rudimentary, limiting our ability to guide management. Methods: We prospectively evaluated a previously established panel of molecular markers in 92 clival chordomas. Our panel included Fluorescensce In Situ Hybridization (FISH) for 1p36 and 9p21 (p16 /INK4A) deletions, as well as microsatellite based Loss Of Heterozygosity (LOH) studies at the following loci: 1p, 9p (p16/CDKN2A), 10q (PTEN), and 17p (TP53). The ki-67 index was also evaluated by immunohistochemistry. Eighty of the analyzed tumors were radiated at some point during follow-up because of residual or recurrence (either with proton beam, radiosurgery, or IMRT). The primary endpoint of freedom from radiographic progression was calculated from the date of surgery/radiation to the latest date of patient follow-up using the Kaplan-Meier method. Molecular markers were assessed for prediction of risk of progression utilizing the Cox Proportional Hazard Model (Stata 14 -StataCorp, College Station, TX). Hazard Ratios (HR) of tumor progression were calculated for every unit increase in the molecular marker variables. Results: After a median follow-up of 45 months, we found increasing total percentages of cells with p16 deletion (HR = 1.02, p < 0.001), increasing percentages of homozygous p16 deletion (HR = 1.03, p < 0.001), and decreasing 9p21/CEP ratios (HR = 0.08, p < 0.001), as assessed by FISH, to strongly predict tumor recurrence and shorter progression-free survival. Heterozygous and complex p16 deletions were not predictive and neither was the rate of monozygosity. Increasing percentages of 1p36 deleted cells (HR = 1.01, p = 0.012), and decreasing 1p36/1q25 ratios (HR = 0.13, p = 0.012) were also predictive of recurrence and a shorter progression free survival, as were increasing ki-67 indices (HR = 1.07, p = 0.001), and increasing hyperploidy rates (HR = 1.02, p = 0.018). In patients treated with radiation, a shorter progression free survival was predicted by increasing percentages of p16 deletion (HR = 1.01, p = 0.014), increasing percentages of homozygous p16 deletion (HR = 1.03, p < 0.001), and decreasing 9p21/CEP ratios (HR = 0.99, p = 0.003), as assessed by FISH. Increasing ki-67 indices were also predictive (HR =1.05, p = 0.001). Notably, higher percentages of 1p36 deletions, and lower 1p36/1q25 ratios were suggestive of an increased risk of progression after radiation, but did not reach statistical significance. None of the microsatellite based LOH indices was found to be predictive of any outcomes in a statistically significant fashion. Conclusion: Our study suggests several molecular markers as promising predictors of clival chordoma overall progression-free survival (homozygous p16 deletions, 1p36 deletions, 9p21/CEP and 1p36/1q25 ratios, ki-67), as well as predictors of progression-free survival after radiation (homozygous p16 deletions, 9p21/CEP ratio, ki-67). All of these markers could prove useful adjuncts to clinical decision-making, guiding the management of these tumors. Further validation of these findings is warranted.