Background: Predictive biomarkers or signature(s) for oesophageal cancer (OC) patients undergoing preoperative therapy could help administration of effective therapy, avoidance of ineffective ones, and establishment new strategies. Since the hedgehog pathway is often upregulated in OC, we examined its transcriptional factor, Gli-1, which confers therapy resistance, we wanted to assess Gli-1 as a predictive biomarker for chemoradiation response and validate it.
Methods: Untreated OC tissues from patients who underwent chemoradiation and surgery were assessed for nuclear Gli-1 by immunohistochemistry and labelling indices (LIs) were correlated with pathologic complete response (pathCR) or <pathCR (resistance) and validated in a unique cohort.
Results: Initial 60 patients formed the discovery set (TDS) and then unique 167 patients formed the validation set (TVS). 16 (27%) patients in TDS and 40 (24%) patients in TVS achieved a pathCR. Nuclear Gli-1 LIs were highly associated with pathCR based on the fitted logistic regression models (P<0.0001) in TDS and TVS. The areas under the curve (AUCs) for receiver-operating characteristics (ROCs) based on a fitted model were 0.813 (fivefold cross validation (0.813) and bootstrap resampling (0.816) for TDS and 0.902 (fivefold cross validation (0.901) and bootstrap resampling (0.902)) for TVS. Our preclinical (including genetic knockdown) studies with FU or radiation resistant cell lines demonstrated that Gli-1 indeed mediates therapy resistance in OC.
Conclusions: Our validated data in OC show that nuclear Gli-1 LIs are predictive of pathCR after chemoradiation with desirable sensitivity and specificity.