| Objective: Olfactory receptor family 13 subfamily C member 2(OR13C2)is an important member of Olfactory receptor family(ORF)whose abnormal expression has been shown in such tumors as hepatocellular carcinoma(HCC)and involved in tumorigenesis and progression of some tumors.However,it is unclear how OR13C2 expresses in HCC and what the corresponding clinicopathological significance is.Methods: HCC patients undergoing surgical resection at the Affiliated Hospital of Youjiang Medical University for Nationalities from August 2017 to March 2022 were collected as the study subjects.The paraffin-embedded tissues samples with HCC were collected from all study subjects and the amount of OR13C2 expression in these samples was tested by immunohistochemistry technique.The associations between OR13C2 expression and clinicopathological features of HCCs were analyzed by chi-square test,Fisher’s exact probability,and Wilcoxon two-related sample rank-sum test,whereas effects of OR13C2 expression on HCC prognosis were analyzed by Kaplan-Meier survival model and Cox proportional hazards regression model.Results:(1)The expression level of OR13C2 in cancer tissues was lower than that in peri-cancer tissues,with a statistically significant difference(Z = 7.052,P < 0.001).(2)According to the median level of OR13C2 expression in cancer tissues,the information of OR13C2 expression was divided into two groups: high and low expression group.Different expressions of OR13C2 were negatively associated with history grades of tumor(correlation coefficient r =-0.213 and corresponding P = 0.023).Furthermore,results from logistic regression analyses showed that decreasing OR13C2 expression increased the risk of tumor dedifferentiation of HCC,with a risk value of 3.16(1.05-9.54)and a P value of 0.023.However,the levels of OR13C2 expression were not correlated with the other clinicopathological features of HCC.(3)The levels of OR13C2 expression in tumor tissues did not affect the prognosis of HCC(P > 0.05).Conclusion: Our results showed that the amount of OR13C2 expression in cancerous tissues decreased compared to peri-cancerous tissues and this decreasing expression increased tumor dedifferentiation of HCC.However,OR13C2 expression did not significantly affect HCC prognosis.These results suggest that the dysregulation of OR13C2 expression may involve in HCC carcinogenesis and play an important role in the tumor dedifferentiation process of tumor.Objective: Hepatocellular carcinoma(HCC)has a high incidence in China,with about half of the world’s liver cancer patients in China.Due to the high degree of malignancy of HCC,the prognosis of patients is often poor,so improving the prognosis of patients has attracted more and more attention from the society.By analyzing preoperative clinical data of patients with liver cancer,comparing clinical characteristics of patients with liver cancer,analyzing adverse prognostic factors and establishing prognostic prediction models,it will be beneficial to provide certain references for individualized adjuvant therapy of patients with liver cancer and improve the prognostic effect of patients with liver cancer.Due to the complex pathogenesis of HCC and many prognostic factors,the prognosis cannot be quantified and the prediction accuracy is not high.Using machine learning to assist prognostic analysis can improve the prediction accuracy and thus improve the cure rate of liver cancer.Therefore,the second part of this study is to use the SEER database to analyze the clinicopathological characteristics and prognostic risk factors of HCC patients,so as to establish an evaluation model with good predictive ability to improve the evaluation ability of individualized patient prognosis.Methods:(1)A total of 1713 hepatocellular carcinoma patients conforming to the natrol criteria were extracted from the SEER database,and their related clinical information was downloaded,including: Survival time,survival state,age,diagnosis year,gender,race,AFP,liver fibrosis score(FS),pathological grade,T stage,N stage,M stage,chemotherapy,surgery and radiotherapy,etc.A total of 1713 patients were randomly divided into two data sets,namely the modeling set and the test set,in a 7:3 ratio.Data of 79 patients with hepatocellular carcinoma were retrospectively collected for an external validation set.(2)Kaplan-Meier method and Cox proportional risk model were used to compare clinical information characteristics and prognostic differences.(3)The 1-,3-,5-and 7-year overall survival prediction models of Cox,random survival forest and gradient lift were established,and the predictive effects of the three models were compared by AUC values.Based on the data of our hospital,the value of different prediction models in the prognosis of HCC patients was explored to provide basis for improving the prognosis.Results:(1)In SEER,1713 HCC patients mainly ranged in age from 60 to 74 years old,with a male to female ratio of 3:1.They are predominantly Caucasian.Most patients had moderately differentiated tumors(55.22%).Most of the patients had TNM stage I to II(86.98%).Local tumor destruction accounted for 22.94% of patients,surgical resection accounted for 52.01%,and liver transplantation accounted for 25.05%.25.74% of patients had definite radiotherapy and 2.98% had definite chemotherapy.They were randomly divided into the modeling set(1199 cases)and the test set(514 cases)at a ratio of 7:3.There were no statistically significant differences between the modeling set and the verification set in age,sex,race,tumor grade,tumor T stage,tumor N stage,tumor size,alpha-fetoprotein,liver fibrosis score,primary surgery,radiotherapy,and chemotherapy(P>0.05).(2)Survival curves of each variable were plotted based on Kaplan-Meier and Log-rank tests.The results showed that the older the HCC patients were,the higher the T,N and M stages of the tumor,the lower the degree of tumor differentiation,and the lower the overall survival rate of patients with lung metastasis and elevated alpha-fetoprotein(P < 0.05).There was no significant difference in overall survival between postoperative chemotherapy patients and those who did not receive chemotherapy(P>0.05).Cox univariate survival analysis showed that age,tumor grade,tumor T stage,tumor N stage,tumor M stage,tumor size,lung metastasis,and alpha fetoprotein were associated with the prognosis of OS in HCC patients(all P < 0.05).Multivariate Cox regression analysis showed that age,tumor grade,tumor T stage,tumor M stage,tumor size,and alpha-fetoprotein were independent factors affecting the prognosis of OS in HCC patients(all P < 0.05).(3)The factors affecting the prognosis of HCC patients were visualized by means of a graph,and the predictive efficacy of the model was evaluated in the modeling set and test set.The ROC curves of modeling set and test set were 0.663,0.666,0.640,0.614 and 0.631,0.639,0.637,0.614 for OS in years 1,3,5 and 7,respectively.(4)ROC curves predicted by random forest model on modeling set showed that AUC of 1-year OS,3-year OS,5-year OS and 7-year OS were 0.747,0.749,0.737 and 0.721.The ROC curve predicted by the test set showed that the AUC of 1-year,3-year,5-year and 7-year OS was 0.645,0.624,0.624 and 0.609.(5)The ROC curve predicted by the gradient lifting model against the modeling set showed that the AUC of 1-year,3-year,5-year and 7-year OS were 0.667,0.663,0.640 and 0.626.The ROC curve predicted by the test set showed that the AUC of 1-year,3-year,5-year and 7-year OS was 0.624,0.601,0.594 and 0.580.(6)After external verification,it was found that the AUC of OS predicted by the line graph model was 0.622,0.597 and 0.558 in the first,third and fifth years,respectively.The random survival forest model predicted that the AUC of 1-year OS was 0.582,0.763 and 0.892,respectively.The gradient lifting tree model predicted that the AUC of 1-year OS was 0.600,0.762 and 0.680,respectively.Conclusion:(1)Cox regression analysis showed that age,tumor grade,tumor T stage,tumor M stage,tumor size,and alpha-fetoprotein were independent factors affecting the prognosis of OS in HCC patients.(2)The risk model was constructed according to the prognostic risk factors.The line graph model,random survival forest model and gradient ascending tree model all had moderate predictive value.(3)The prediction effect of the random survival forest model is better than that of the line graph model and the gradient lifting tree model,and the survival rate predicted by the survival forest model is closer to the actual survival rate of patients. |
PDF Full Text Request