Preliminary Study Of CT Radiomics Analysis On Differentiating Exon 9/11 Mutations Of C-kit Gene In Gastrointestinal Stromal Tumors

Gastrointestinal stromal tumor(GIST)is the most common mesenchymal tumor of gastrointestinal tract,and the stomach is the best site to exit.Type is80% ～ 85% of GIST for Ⅲ tyrosine kinase(c-kit)caused by gene mutations,in the study of kit gene mutation of c-kit found that about 70% of the mutation in exon 11 GIST,exon 9 mutation is 5% ～ 10%.Surgical resection is the preferred treatment in the clinic,but molecular targeted drug therapy is needed for GIST that is unrespectable,relapsed or metastatic,and at medium and high risk.However,it is worth noting that the drug selection,dosage and efficacy of targeted therapy are closely related to the mutated gene type of GIST.According to China’s consensus recommendation on the diagnosis and treatment of gastrointestinal stromal tumors,individual gene testing is required for GIST patients before targeted therapy,which is of great significance for the development of personalized treatment plans and efficacy evaluation.However,due to the high cost,routine genotyping of patients for GIST is still not widely used.Abdominal CT enhanced examination is the most common examination method for GIST patients in clinical practice,which has a high application value in the diagnosis of GIST.Radiomics extracts and analyzes a large number of high-throughput imaging features from conventional medical imaging images,and decodes the subtle relationship between clinical and gene mutations,so as to be used in disease diagnosis,efficacy evaluation and prognosis prediction.CT texture analysis technique(CCTA),through the use of special algorithm to extract a series based on pixel space distribution,the human eye can’t identify the quantitative texture parameters,such as the entropy value,the value of skewness,peak,to deep mining the data information of the original image,which can be more comprehensive,meticulous to reflect the characteristics of the lesions,provides a medium for the realization of the Radiomcis.It is a research hotspot in the field of cancer therapy to explore the correlation between image features and genes through the method of imaging omics,which provides guidance for clinical targeted therapy and prognosis evaluation.Objective:To explore the use of CT imaging omics analysis to identify exon 9 and11 mutations of c-kit gene,and preliminarily build a prediction model capable of classification diagnosis,providing more basis for intelligent and personalized treatment of GIST.Methods:This study retrospectively collected in the center of Baoding city first hospital from 2018 to 2020 and the fourth hospital,Hebei medical university by spiral CT detection,and 49 patients with confirmed genetic pathology as GIST,and 9/11 of kit gene c-extra child have a mutation,and the related clinical pathologic data collection,all patients with preoperative no systemic treatment(chemotherapy,targeted therapy,etc.).All patients were scanned by Brilliance ICT 256-slice spiral CT with three-phase enhanced scanning.The images of arterial phase,portal phase and delay phase were obtained.The arterial phase images were selected,ITK-SNAP 3.8.0 was used to outline the region of interest,and Python Py Radiomics 3.0.1 was used to extract the CT Radiomics features.The selected omics features with the most correlation were included in the next statistical analysis.Finally,the logistic Regression(LR)algorithm was used to construct the gene identification model.SPSS 22.0 statistical software was used for relevant statistical analysis,and Mann-Whitney U test was used to evaluate the differences of various clinical factors and imaging features among the 9/11 exon mutations.P < 0.05 was considered to be statistically significant.The difference of the final omics characteristics between the two groups of data was shown through boxplot.Receiver Operating Characteristic Curve(ROC)and area under Curve(AUC)were used to evaluate the characteristics of each group and the diagnostic performance of the differential model.Results:1.To investigate the relationship between two exon mutations of c-kit gene 9/11 and clinicopathological data of GIST patientsIn the 49 GIST patients who were finally enrolled,gender and age did not have statistical significance to distinguish the 9/11 exon mutations of c-kit gene(P>0.05).2.To explore the correlation between CT imaging omics analysis and exon 9 and 11 mutations of c-kit gene(1)based on omics feature extraction to the 105 image,after a LASSO feature selection(as shown in figure 1),finally chosen seven strong correlation characteristics,Small High Gray correlation respectively stress factor(Small Dependence High Gray Level Emphasis,SDHGLE),clustering shadow(Cluster Shade),the largest 3 d(Maximum 3 d Diameter,M3D)in Diameter,Skewness(Skewness),intensity(Strength),Maximum 2D Diameter Slice(M2DS)is Elongation.M3 D and M2 DS were selected by Lasso and showed a high correlation(R=0.96).SDHGLE,Skewness,Strength,and Elongation were significantly higher in the 9 mutation than in the 11 mutation(P<0.05),while M3 D and M2 DS were significantly lower in the 9 mutation than in the 11 mutation(P<0.05).Cluster Shade was not statistically different between the two groups(P>0.05).(3)The Radiomics features and the final omics model showed superior diagnostic performance in the ROC curve for identifying the two gene mutations,with AUC of 0.69,0.68,and 0.67,respectively.The final logistic regression model has the best diagnostic performance,with an AUC of 0.79.Conclusion:The results showed that SDHGLE,Skewness,Strength,and Elongation had higher expression in GIST patients with c-kit exon 9 mutation.M3 D and M2 DS were lower in patients with exon 11 mutations.In a preliminary study,CT imaging showed some predictive potential in identifying mutations in the9/11 exon of c-kit gene in gastrointestinal stromal tumors.Therefore,combined imaging technology may provide reference for the identification of c-kit gene mutation types in GIST,and thus provide guidance for the individualized targeted therapy of GIST.