Identification Of SncRNA Biomarkers And Construction Of Models For Diagnosis,Lymph Node Metastasis Prediction And Prognosis Of Colorectal Cancer

BackgroundColorectal cancer(CRC)is one of the most frequently diagnosed malignancies worldwide with substantial mortality.The high degree of mortality is largely due to its late detection and high recurrence rates,and the 5-year survival rate of patients is highly dependent on the cancer stage at diagnosis.Early detection is thus crucial for effective treatment of CRC and patient survival.Lymph node(LN)metastasis is also an important determining factor for the outcome of CRC.Accurate preoperative prediction of LN status in CRC is of crucial importance for appropriate therapeutic decisions.However,conventional tests,such as imaging modalities and colonoscopy,usually have several disadvantages including low sensitivity and specificity,high cost or discomfort to the patients.Therefore,it is urgently necessary to develop sensitive,specific,cost-effective and non-invasive methods to complement and improve current screening strategies for CRC diagnosis and preoperative prediction of LN metastasis in CRC patients.Additionally,the most commonly used approach for predicting patient survival remains pathological staging according to the tumor-node-metastasis(TNM)classification system,but it provides only limited information for the clinical prognostication since even patients within the same stage display a strong heterogeneity for prognosis and treatment response.This highlights the critical need to develop robust prognostic biomarkers that can offer a superior prognostic clinical usefulness compared with the traditional TNM staging system.Recent advancements in transcriptome profiling have highlighted the potential of small non-coding RNAs(sncRNAs)as tumor biomarkers.Piwi-interacting RNAs(piRNAs)are the largest class of sncRNAs,which have been of interest primarily due to their essential role in the maintenance of germ cells.Recent studies have shown that some piRNAs are involved in carcinogenesis of several types of cancers,including CRC,providing compelling evidence for the role of these piRNAs in cancer development.Moreover,it has been reported that the piRNAs present in human plasma and serum from both healthy individuals and cancer patients,in a markedly stable form.These studies imply that the aberrant expression of piRNAs may be detected in easily obtained bodily fluids from cancer patients.However,the expressions of circulating piRNAs in cancer patients and their potential clinical utilities remain largely unexplored.Another class of small ncRNAs,microRNAs(miRNAs),are the best characterized and regulate gene expression post-transcriptionally in multiple cancer-related processes including metastasis.Accumulating evidence has provided insight into the role of dysregulated miRNAs as potential tumor markers to predict disease progression and metastasis.Recently,differential expressions of circulating miRNAs have been reported to predict LN metastasis in various cancers.Although many studies have proposed circulating miRNAs to be predictors of metastasis,very few have attempted to identify circulating miRNA-based panel for prediction of LN status before surgery.Additionally,hypoxia is a poor-prognosis microenvironmental hallmark of most solid tumours.And hypoxia-responsive miRNAs also play critical roles in human cancers,which are closely related to poor prognosis of patients.However,to date,there is no comprehensive analysis of predictive biomarkers for prognosis based on hypoxia-induced dysregulated miRNA expression profiles in CRC patients.In the present study,we systematically and comprehensively screened CRC-specific piRNA biomarkers and LN metastasis-specific miRNAs in the serum samples from CRC patients,respectively.Then we established a non-invasive piRNA-based panel for CRC diagnosis,and developed an inclusive nomogram incorporating an miRNA panel and CT-reported LN status for the prediction of LN metastasis in CRC patients.Additionally,we identified hypoxia-and survival-associated miRNAs,and further constructed a prognostic model for predicting the prognosis in patients with CRC.Our study will help improve the non-invasive diagnosis,preoperative prediction of LN metastasis and prognosis prediction of CRC patients,and is expected to contribute to optimize individually-tailored therapy in CRC.Part One.Identification of serum piRNA biomarkers and construction of a diagnostic model for CRCOBJECTIVE:Piwi-interacting RNAs(piRNAs)are a novel class of small non-coding RNAs,which are not easily degraded but detectable in human body fluids.In the present study,we systematically and comprehensively screened CRC-specific piRNA biomarkers in the serum samples,and established a non-invasive piRNA-based panel for CRC diagnosis.METHODS:1.In the screening phase,serum samples pooled from ten patients with TNM ?-?stage(CRC Group A),ten patients with TNM III-IV stage(CRC Group B)and ten healthy controls were subjected to sequencing to identify piRNAs that were significantly differentially expressed.2.In the training phase,we firstly tested the expression of candidate piRNAs in a cohort of 120 CRC patients and 120 controls using RT-qPCR assay,and then constructed a piRNA-based diagnostic panel using logistic regression model.Receiver operating characteristic(ROC)curve was established to examine the diagnostic value of the panel for CRC.3.In the validation phase,we evaluated the diagnostic performance of the piRNA-based panel in another independent cohort of 100 CRC patients and 100 healthy controls using ROC curve analysis.4.We further investigated the serum levels of CEA and CA19-9,and constructed CEA-based panel(including CEA and CA19-9)using logistic regression model.A comparison between piRNA-'based panel and CEA-based panel was performed.Also,Fagan's nomogram was plotted to promote the clinical application of piRNA-based panel.5.A total of 110 patients with complete medical records and adequate follow-up from both training and validation sets were included in the survival analysis.Kaplan-Meier analysis and Cox proportional hazard regression model were used to analyze the correlation between serum piRNAs and prognosis in CRC patients.RESULTS:1.Among the 23,439 serum piRNAs scanned by Illumina HTS,there were 16 piRNAs that were differentially expressed between the CRC patients and healthy controls.Thus,this 16 piRNAs were selected as candidates for further analysis.2.In the training phase,RT-qPCR analysis revealed that five piRNAs(piR-001311,piR-004153,piR-017723,piR-017724 and piR-020365)displayed differential expression patterns between CRC patients and controls(all P<0.05).ROC analysis showed that the corresponding AUCs of these five piRNAs were 0.726,0.729,0.756,0.749 and 0.786,respectively.3.A diagnostic piRNA panel(piRNA-based Panel 1)based on the five piRNAs for CRC was constructed by logistic regression model as follows:Logit(P=CRC)=-1.0391 + 0.1621 x piR-001311 + 0.203 × piR-004153 + 0.3711 × piR-017723 +0.0897 x piR-017724 + 0.128 ×piR-020365.ROC curve analysis revealed that the AUC for the five-piRNA panel was 0.867(95%CI:0.817-0.907)with a sensitivity of 78.3%and a specificity of 74.2%(cut-off value =-0.1085).4.The piRNA-based Panel I obtained from the training set was used to predict the probability of being diagnosed with CRC in the validation set.ROC curve analysis revealed that the AUC of piRNA-based Panel I was 0.854(95%CI:0.797-0.900).The sensitivity and specificity were 78.0%and 76.0%,respectively,by setting-0.1085(derived from training set)as the threshold value.5.In the training set,the AUC of the conventional markers CEA and CA19-9 were 0.718 and 0.571,respectively.Using logistic regression model,we constructed CEA-based Panel II(including CEA and CA19-9)and plotted its ROC curve with an AUC of 0.735(95%CI:0.675-0.790).Moreover,we compared the AUC values of piRNA-based Panel I and CEA-based Panel ?,and the results showed that Panel I had significantly higher AUC value than Panel ?(P<0.001).6.In Fagan's nomogram,we set the pre-test probability of CRC at a hypothetical value of 5%.The positive likelihood ratio(+LR)of panel I was 3.13 and the negative likelihood ratio(-LR)was 0.29.If a patient had a positive piRNA-based Panel I result,the post-test probability that he truly has CRC would rise to 14.1%.Alternatively,if the patient had a negative result,the post-test probability would decrease to 1.5%.7.Kaplan-Meier survival analysis showed that patients with low expression of piR-017724 had significantly lower overall survival(OS)and progression-free survival(PFS)rate compared with those with high expression of piR-017724(P=0.005 and P=0.002,respectively).Cox proportional hazards model showed that piR-017724 expression was an independent prognostic factor for OS and PFS in CRC patients(P=0.003 and P=0.004,respectively).CONCLUSIONS:1.In the present study,we for the first time demonstrated a potentially clinically applicable serum piRNA panel(piR-001311,piR-004153,piR-017723,piR-017724 and piR-020365)for tumor detection and meanwhile,serum piR-017724 was identified as an independent prognosis predictor in patients with CRC.2.The diagnostic potential of this five-piRNA based panel was better than that of conventional biomarkers.Furthermore,we constructed a Fagan's nomogram which could facilitate the interpretation of test results into clinically useful information.3.Our study provided a new type of biomarkers with great applicability potential and represented a new approach for the non-invasive diagnosis of CRC.Part Two.Development of a preoperative prediction nomogram for lymph node metastasis in CRC based on a novel serum miRNA panel and CT scansOBJECTIVE:Preoperative prediction of lymph node(LN)status is of crucial importance for appropriate treatment planning in CRC patients,which is also an urgent clinical problem to be solved.In this study,we sought to develop and validate a non-invasive nomogram model by systematically and comprehensively screening LN metastasis-specific miRNAs,to preoperatively predict LN metastasis in CRC patients.METHODS:1.In the discovery stage,serum samples pooled from ten patients without LN metastasis(LN-group)and ten with LN metastasis(LN+ group)were subjected to high-throughput sequencing(HTS)to identify differentially expressed miRNAs.A miRNA was considered "significantly altered" only if the RPM value was larger than 1,along with a larger than two-fold change in its expression level in the LN-group vs.the LN+ group.2.In the training stage,candidate miRNAs were firstly examined by RT-qPCR in serum samples from 108 LN-and 110 LN+ patients(training set).Subsequently,differentially expressed miRNAs were further confirmed in another cohort of 98 LN-and 100 LN+ patients(the validation set).3.A miRNA-based panel for LN status prediction was constructed using the logistic regression model in the training set based on the differentially expressed miRNAs.ROC curve and corresponding AUC were employed to assess the performance of miRNA-based panel for LN status prediction in both training and validation sets.4.We used a multivariable logistic regression model to identify the preoperative clinical risk factors that were significantly correlated with LN status and then combined them with the miRNA-based panel to construct an inclusive nomogram model in the training set.The Hosmer-Lemeshow test was used to examine the goodness-of-fit of the nomogram model.ROC curve,calibration curve and decision curve analysis(DCA)were employed to assess the discriminative ability,consistency and clinical utility of the nomogram,respectively.RESULTS:1.In the discovery stage,317 and 351 miRNAs were detectable in LN-group and LN+ group,respectively.Among them,30 candidate miRNAs were identified to be differentially expressed between LN-group and LN+ group.2.RT-qPCR analysis demonstrated there were four differently expressed miRNAs(miR-122-5p,miR-146b-5p,miR-186-5p and miR-193a-5p)between LN-and LN+patients in the training set(all P<0.05).Among them,miR-122-5p,miR-146b-5p and miR-186-5p were upregulated in LN+ patients,while miR-193a-5p was downregulated.The AUC values of these four miRNAs for LN prediction were 0.726(95%CI:0.662-0.784),0.811(95%CI:0.753-0.861),0.744(95%CI:0.681-0.800)and 0.681(95%CI:0.615-0.742).The alteration patterns of the miRNA expression in the validation set were consistent with those in the training set,with AUCs ranging from 0.722 to 0.796.3.Through logistic regression analysis,a risk score formula of miRNA-based panel was built to predict LN status as follows:Logit(P=LN metastasis)=-1.916 +miR-122-5px0.495 + miR-146-5p×0.869 + miR-186-59×0.899 + miR-193a-5p×(-0.377).ROC analysis showed that the 4-miRNA panel yielded an AUC of 0.907(95%CI:0.860-0.942),with a sensitivity of 88.2%and a specificity of 76.4%(cut-off value =-0.63).4.The 4-miRNA panel obtained from the training set was used to predict LN status in the validation set.ROC curve analysis revealed that the AUC of 4-miRNA panel was 0.870(95%CI:0.815-0.913).The sensitivity and specificity were 78.0%and 83.0%,respectively,by setting-0.63(derived from training set)as the threshold value.The AUC for CT-reported LN status were 0.623(95%CI:0.549-0.697)in the training set and 0.675(95%CI:0.600-0.750)in the validation set,which were both significantly lower than that of miRNA-based panel(both P<0.05).Moreover,the miRNA panel showed a good discriminatory ability in the CT-reported LN negative subgroup,with AUC values of 0.797(95%CI:0.701-0.892)in the training set and 0.764(95%CI:0.660-0.868)in the validation sets.5.Logistic regression analysis revealed the 4-miRNA panel(OR:2.605,95%CI:1.272-5.335,P=0.009)and the CT-reported LN status(OR:22.902,95%CI:10.819-48.478,P<0.001)were both independent risk predictors of LN status.A nomogram model that incorporated the above two independent predictors was developed in training set.The nonsignificant Hosmer-Lemeshow test statistic(P=0.268)indicated a good fit to the nomogram model.The calibration plot of our nomogram showed the bias-corrected line lay close to the ideal curve(the 45-degree line),implying a good agreement between prediction and observation.ROC analysis yielded an AUC of 0.913(95%CI:0.878-0.948),which implied the discrimination performance was favorable.DCA showed that using the nomogram to predict LN status adds more benefit than the "treat-all" or "treat-none" scheme.It also showed higher net benefit than CT-reported LN status.CONCLUSIONS:1.In the present study,we identified for the first time a novel serum 4-miRNA panel(miR-122-5p,miR-146b-5p,miR-186-5p and miR-193a-5p)that discriminated with high accuracy between the serum miRNA profiles of CRC patients with and without LN metastasis.2.An inclusive nomogram incorporating the 4-miRNA panel and CT-reported LN status was constructed for the prediction of LN metastasis in CRC patients,which displayed satisfactory predictive accuracy in both the training and validation sets.3.This predictive nomogram model has great applicability potential in the non-invasive clinical evaluation of patients at risk of LN metastasis,and may be conveniently used to optimize treatment strategies by avoiding unnecessary LN-related procedures in CRC.Part Three.Prognostic and predictive value of a hypoxia-miRNA panel in colorectal cancerOBJECTIVE:Hypoxia is a poor-prognosis microenvironmental hallmark of most solid tumours.Hypoxia-responsive miRNAs also play critical roles in human cancers,which are closely related to poor prognosis of patients.In the present study,we aimed to identify hypoxia-and survival-associated miRNAs,and to further construct a prognostic model for predicting the prognosis in patients with CRC.METHODS:1.In the discovery stage,HT-29 cells under normoxic and hypoxic conditions were subjected to miRNA sequencing to identify differentially expressed miRNAs.The expression of a miRNA was considered altered only if at least 50 counts were detected by HTS,together with fold change>1.5 in its expression level between the hypoxia cells and normal ones.2.In the training stage,the candidate hypoxia-induced miRNAs were initially subjected to univariate Cox proportional hazard regression analysis to examine the association between miRNA expression and overall survival.The miRNAs with top statistical significance were subsequently entered a step multivariate Cox regression analysis to train a hypoxia-induced miRNA panel.Kaplan-Meier curve and ROC curve were used to assess the prognostic performance of the miRNA-based panel.3.According to the obtained miRNA-based panel,we computed the riskscore for all patients in the testing and validation sets,and divided them into high-risk group and low-risk group using the same cutoff value obtained from the training set.We did the same survival analysis(Kaplan-Meier curve and ROC curve)to examine the prognostic value of the miRNA panel in both testing and validation sets.4.In the entire cohort of 791 patients(combination of the training,testing and validation sets),Cox regression analysis and Kaplan-Meier stratification analysis were performed to investigate whether the prognostic value of the miRNA panel was independent of other clinicopathological variables.We also performed ROC analysis to compare the predictive ability of the miRNA panel with other features.5.A nomogram was constructed which integrated both the miRNA panel and clinicopathological independent risk factors for patients' survival.Calibration curve and ROC curve were employed to assess the consistency and discriminative ability of the nomogram,respectively.RESULTS:1.There were 66 differentially expressed miRNAs in CRC cells under hypoxia conditions when compared to normoxic controls.Of these,52 miRNAs were upregulated by hypoxia and were considered to be candidate hypoxic miRNAs and subj ected to subsequent analysis.2.Among the 52 miRNAs subj ected to univariate Cox proportional hazards regression analysis in the training set,eight candidates with top statistical significance(P<0.1)were subsequently entered into a step multivariate Cox regression analysis.As a result,we trained a hypoxia-related prognostic panel as follows:Riskscore(0.2113×miR-210)+(0.4688×miR-26a)+(0.4337×miR-197)+(-0.2266×miR-375).The patients were dichotomized into high-risk group and low-rnsk group according to the optimum cutoff value generated by X-tile plots(Riskscore=1.63).Kaplan-Meier survival analysis revealed high risk patients had shorter OS than low risk patients(log-rank test,P<0.001).ROC analysis showed that the AUC for the four-miRNA prognostic model was 0.711(95%Cl:0.630-0.791)at 3 years and 0.737(95%CI:0.627-0.845)at 5 years.3.In line with the findings of training set,high risk patients had shorter OS than low risk patients in the testing set(log-rank test,P=0.042).Time-dependent ROC analysis indicated the AUC for the miRNA panel was 0.568(95%Cl:0.500-0.635)at 3 years and 0.657(95%Cl:0.518-0.797)at 5 years.The results of validation set showed that more patients with dead status fell into the high-risk group,in which the OS was shorter than that in low-risk group(P<0.001);the miRNA panel achieved AUC of 0.688(95%CI:0.598-0.778)and 0.686(95%CI:0.539-0.833)at 3 and 5 years.4.Univariable and multivariable Cox regression analyses showed that patient's age(HR=3.204,95%CI:1.910-5.374,P<0.001),preoperative CEA level(HR=2.166,95%CI:1.392-3.371,P=0.001),TNM stage(HR=1.944,95%CI:1.233-3.065,P=0.004)and the miRNA panel(HR=2.806,95%CI:1.724-4.566,P<0.001)were independent prognostic factors in the entire cohort of 791 patients(combination of the training,testing and validation sets.Stratification analysis according to age,preoperative CEA level and clinical stages showed that high risk patients had significantly shorter OS than low risk patients in all subgroups except for younger group(all P<0.05).In younger subgroup,the difference was marginally significant(P=0.088).5.The time-dependent ROC curve analysis was performed with 3 years and 5 years as the cut-off point.The results showed that the AUC of the miRNA panel(0.667)was significantly larger than that of TNM stage(0.616),age(0.579),CEA(0.624)and single miRNA(miR-375:0.542,miR-197:0.617,miR-26a:0.550,miR-210:0.579)(all P<0.05).Similarly,at the cut-off point of 5 years,the AUC of the miRNA panel(0.704)was significantly larger than that of TNM stage(0.614),age(0.608),CEA(0.635)and single miRNA(miR-375:0.588,miR-197:0.616,miR-26a:0.558,miR-210:0.569)(all P<0.05).This suggests that the four-miRNA risk score model possessed a stronger predictive power than any other clinical risk factors or single miRNA alone.6.A nomogram was constructed which integrated the miRNA panel and clinicopathological independent risk factors for patients' survival(including age,TNM stage and preoperative CEA level).The bias-corrected lines of both 3 and 5 years in the calibration plot were very close to the ideal curve(the 45-degree line),which indicated good agreements between prediction and observation.ROC analysis showed that the AUCs of nomogram at 3 and 5 years were 0.763(95%CI:0.704-0.822)and 0.752(95%CI:0.678-0.827)respectively,which demonstrated the favorable discrimination performance.CONCLUSIONS:1.In the present study,we for the first time constructed a novel hypoxia-related miRNA-based panel that could predict individualized survival for CRC patients.2.This novel hypoxia-related miRNA panel was an independent prognostic factor of,and possessed stronger predictive power than,currently used clinicopathological features for identifying high-risk CRC patients3.An inclusive nomogram incorporating the hypoxia-related miRNA panel and independent prognostic factors was constructed,which could be a useful tool for patient counselling and personalize management for patients with CRC.