Development Of Predicting Models For Craniopharyngioma Resection Extent/Recurrence And The Mechanism Of Transcription Factor

BackgroundCraniopharyngiomas(CPs)are benign suprasellar tumors accounting for 2-4%of intracranial tumors.There are two types of CPs,the adamantinomatous(ACP)and the papillary(PCP)based on the pathological type.Although histologically classified as WHO I tumors,total resection and postoperative management are huge challenges associated with CPs because they are adjacent to vital brain structures,such as the optic chiasm,hypothalamus as well as anterior communicating artery complex.Radical resection is considered the first-line treatment because it yields the best overall survival(OS)and progression-free survival(PFS).However,complete resection could lead to increased mortality or poor functional results because of severe endocrine disorders.The rate of long-term recurrence could reach an astonishing 58%with or without radiotherapy in the subtotal resection group of some studies,so subsequent treatment often becomes an unavoidable problem for such patients.Secondary resection,salvage radiotherapy or intratumoral chemotherapy can be carried out in relapsed patients,though its therapeutic effect is not satisfactory.In the long-term follow-up,clinical doctors generally pay more attention to the endocrine status of patients than to the neuropsychological status.Only a few studies have focused on these issues.Nevertheless,there is still a lack of detailed preoperative and postoperative neurological and neurological evaluations in a large cohort.In recent years,an increasing number of studies have focused on baseline clinical characteristics,such as image features and laboratory tests to predict prognosis and long-term disease recurrence.Nomograms,as more intuitive and visual tools,have been increasingly used in clinical practice.For instance,the Memorial Sloan Kettering Cancer Center(MSKCC)launched different predictive models to guide clinical treatment more accurately.Although nomograms are extensively used in clinical work,due to the low morbidity associated with CPs,predictive models for CP have not yet been established.Long noncoding RNA(LncRNA)are a type of RNA with lengths over 200 nucleotides that are not translated into protein.Mounting researches suggested LncRNAs exert impact on numerous biological process,such as cell proliferation,invasion,differentiation,apoptosis and metastasis.Similarly,it has been suggested transcription factors(TFs)modulated the expression of LncRNA to mediate downstream molecule expression and promote tumor development.At present,there are few studies on transcription level in craniopharyngioma,but the research on post-transcriptional regulation,especially LncRNA,has not been reported.Objective1.This study aimed to develop a novel prognostic model for CP combining preoperative and postoperative features based on three large centers.The long-term neuropsychological status of patients was also assessed with the QLQ-BN20 questionnaire.2.Based on the high-throughput RNA sequencing result,we used bio-informatics analysis to investigate the dataset of ACP and tumor tissues were experimentally validated to explore the relationship between gene expression and the tumorigenesis of ACP.Method1.A total of 545 consecutive patients were diagnosed with CPs and treated at the Department of Neurosurgery,the First Affiliated Hospital of Zhengzhou University,Henan Provincial People’s Hospital and the Third Affiliated Hospital of Zhengzhou University between October 2009 and October 2019.Based on these criteria,381 patients were included in the development cohort,and 164 patients(7:3)were selected by random extraction and included in the validation cohort.All the clinical parameters of patients were recorded and obtained from the electronic medical system.EORTCQLQ-BN20 questionnaire was used to evaluate the neuropsychological state of patients.2.Least absolute shrinkage and selection operator(LASSO)and Cox regression analyses were performed to establish two nomograms.Receiver operating characteristic(ROC)curves,calibration curves,decision curve analysis(DCA)and Kaplan-Meier(KM)curves were used to evaluate their predictive performance and discriminative power,respectively,in the two cohorts.We further categorized patients into high-and low-risk groups based on their median risk score.Finally,a risk factor-stratified calculator for long-term recurrence was generated and is available in the additional information online.3.Transcriptome level sequencing were detected in 12 human ACP and 5 control samples.We built an integrated algorithm for identifying LncRNA and transcription factor regulators of craniopharyngioma-related pathway.Further,the ChIP-Seq datasets with binding domain information were used to further verify and identify the TF-LncRNA correlation.RT-PCR and immunohistochemistry staining were applied in the validation of potential targets.Results1.Based on the LASSO and Cox regression models,two nomograms that integrated significant predictors were generated.The nomograms used to predict the extent of resection compromised tumor size,duration of symptoms,hypothalamus involvement,calcification and characteristics.As shown in Fig 2,the AUCs of individual tumor size,duration of symptoms,hypothalamus involvement,calcification and characteristics were 0.587,0.502,0.489,0.244 and 0.422,respectively,but the combination nomogram reached an AUC of 0.760(95%CI:0.73-0.78),with a C-index of 0.758(95%CI:0.721-0.793).To further verify the efficacy of the nomogram,we investigated the model with the validation cohort.The AUC of the external cohort was 0.704(95%CI:0.68-0.74),with an AUC that was obviously higher than that of other individual predictors.The calibration curve yielded agreeable results.DCA showed more favorable clinical applications with the nomograms than with individual predictors,demonstrating the feasibility of the nomograms for making valuable judgments on prognosis.2.To better predict long-term recurrence during follow-up,nomograms involving age,tumor size,hypothalamus involvement,calcification and subtotal resection were generated.The AUCs of the 3-and 5-year PFS nomograms were 0.78(95%CI:0.72-0.83)and 0.75(95%CI:0.69-0.80,C-index:0.785,95%CI:0.757-0.809),respectively;the AUCs in the validation cohort reached 0.72(95%CI:0.69-0.76)and 0.69(95%CI:0.65-0.72,C-index:0.735,95%CI:0.703-0.75),respectively.The calibration curve was favorable.Moreover,DCA was performed in both cohorts.3.5 pathways associated with ACP were identified and defined by extensive literature search.Based on the specific pathways,266 ACP-related LncRNAs,39 TFs were calculated by our integrating algorithm.Comprehensive analysis of ChIP-Seq datasets revealed that 29 TFs targeted with 12000 LncRNAs from a wide range of tissues including 161 ACP-related LncRNAs identified by the computational method.These 29 TFs and 161 LncRNAs constituting 1004 TF-LncRNA pairs played role in the different ACP-related pathways as potential regulators.Validation of RT-PCR and immunohistochemistry staining revealed that the expression of ACP-related TFs,E2F2 and KLF5 were positive in ACP.Moreover,the level of LncRNA RP11-360P21.2 was increased in ACP tissues.On the basis of the above results,we finally construct the network of KLF5-RP11-360P21.2、E2F2-RP11-360P21.2.Conclusion1.In conclusion,these novel nomograms of craniopharyngiomas could better predict the extent of resection before surgery and the risk of long-term recurrence;thus,these nomograms might allow neurosurgeons and patients to benefit from clinical care and decision making.According to risk score,patients were stratified into different risk groups to increase the utility of follow-up surveillance and management strategies and will provide guidance in individual treatment and clinical applications.2.This study identified the potential LncRNAs and TFs and established TF-LncRNA regulatory network in ACP at the posttranscriptional level through RNA sequencing and further clarified the regulation of ACPs at the posttranscriptional level.Targeting the network of KLF5-RP11-360P21.2 and E2F2-RP11-360P21.2 may be a novel therapeutic strategy for ACP in the future.All these results suggested that identification of the critical LncRNAs/or TFs involved in ACP constitutes a valuable resource in the development of precision medicine.