Development Of Prediction Model Of Conditional Survival Rate For Patients With Resected Stage Ⅰ-ⅢA Non-small Cell Lung Cancer

First Part.Development of prediction model of conditional survival rate for patients with resected stage Ⅰ-ⅢA non-small cell lung cancerObjective:There is currently no consensus on the optimal postoperative follow-up strategy for patients with non-small cell lung cancer(NSCLC).This study aims to establish a webbased prediction model to predict the conditional probability of cancer-specific survival(CPCS)of postoperative patients with stage Ⅰ-ⅢA NSCLC,and assist in the development of individual follow-up plans.Methods:We extracted information of 33,241 postoperative patients with stage Ⅰ-ⅢANSCLC from the Surveillance,Epidemiology,and End Results Database(SEER).They are randomly divided into training cohort and internal validation cohort.The inclusion criteria included:(Ⅰ)The histopathological diagnosis was primary single NSCLC;(ⅱ)receiving lobectomy(including surgical procedures involving excision beyond lobe,such as pneumonectomy),sublobar resection(segmentactomy or wedge resection);(ⅲ)Survival>1 month.Exclusion criteria included:a history of other malignancies;lack of detailed or complete information.Variables included race,sex,age,tumor characteristics(location,lateral,histological type,pathological grade),number of dissected lymph nodes,type of surgery,radiotherapy record,chemotherapy record,SEER cause-specific death classification,and survival time(months).The age of patients were categorized into 3 groups namely:<60,60-69,and≥70 years,whereas the number of examined lymph nodes was classified into<16 and≥16.In the training cohort,the variables affecting the prognosis of stage Ⅰ and Ⅱ-ⅢA NSCLC were screened by univariate Cox regression analysis,and the variables significantly affecting the prognosis were included in the web-based CPCS prediction model of stage Ⅰand Ⅱ-ⅢA respectively.The efficacy of the prediction model was verified by an internal validation cohort and the CMU cohort of 347 patients with completely resected NSCLC in the First Affiliated Hospital of China Medical University,respectively.The discrimination and calibration of the nomograms were evaluated using concordance index(C-index)and calibration plots.The net benefits of the nomograms were quantified using decision curve analysis(DCA).Further,we included 299 patients with stage Ⅰ-ⅢANSCLC who underwent surgery at the Cancer Hospital of the Chinese Academy of Medical Sciences as the NCC cohort to evaluate the effectiveness of the predictive model for relapse risk stratification.Results:The median follow-up of SEER cohort and CMU cohort was 43 months(range 1155 months)and 44 months(range 1-143 months)respectively,and the 5-year accumulative cancer-specific survival(ACS)was 72.8%and 68.6%,respectively.The hazard rate of death(HR)increased over time,with the SEER cohort and CMU cohort peaking at 18 months(HR=0.0064)and 14 months(HR=0.0108),respectively,and then decreasing over time.Accordingly,taking 3-year CPCS as an example,the 3-year CPCS of SEER cohort gradually increased from 80.9%in 0 year to 89.6%in 5th year,while ACS gradually decreased from 100%in 0 year to 65.2%in 8th year.The stage I training cohort,internal validation cohort,and CMU cohort contained 14293,9531,and 159 patients,respectively,while the stage ⅡAⅢA training cohort,internal validation cohort,and CMU cohort contained 5650,3767 and 188 patients,respectively.The calibration curves of the prediction models showed good agreement between predicted cancer-specific survival(CSS)and actual CSS.Moreover,the C-index of the development cohort,internal validation cohort and external validation cohort for stage Ⅰ NSCLC were 0.67,0.67,0.77,respectively.The C-index of the development cohort,internal validation cohort and external validation cohort for stage Ⅱ-ⅢA NSCLC were 0.62,0.64,0.69,respectively.DCA showed that the prediction model was clinically useful.There were 299 patients in the NCC cohort with a median follow-up of 37 months(range 1-102 months)and a 5-year relapse-free survival rate of 66.4%,among whose 212 Ⅰ stage patients and 87 Ⅱ-ⅢA stage patients.The recurrence risk showed a unimodal pattern,peaking at 8 months after surgery and then decreasing over time,reaching a secondary peak at about 50 months after surgery.After the NCC cohort was divided into the high-risk,medium-risk and low-risk groups according to the three-year CPCS,the number of cases in the high-risk,medium-risk and low-risk groups was 98.172 and 29.respectively,with 5,60 and 10 recurrent cases,respectively.The risk of recurrence in the low-risk group was always low,the risk of recurrence in the medium-risk group showed a significant bimodal pattern,and most of the recurrence events in the high-risk group occurred in the first two years after surgery.The prediction model can distinguish the patients with low recurrence risk from those with medium and high recurrence risk.Conclusions:This study constructed a user-friendly web-based prediction model for the first time,which can be used for individualized evaluation of CPCS in patients with resected NSCLC,with good discrimination,calibration and clinical practicability,and can effectively distinguish the population with low recurrence risk from those with high recurrence risk.This model may help oncologists assess cancer-specific dynamic survival and develop individualized strategies for postoperative follow-up,especially for long-term survivors.Second Part.SHC1 expression and its partial methylation sites are potential biomarkers that could be used to evaluate the prognosis of LUADObjective:SHC1(SHC adaptor protein 1)is a scaffold protein.The purpose of this study was to explore the potential of SHC1 and its methylation sites as biomarkers to predict prognosis of NSCLC,with a view to using it in combination with predictive models to improve the accuracy and specificity of the model.Methods:We used Oncomine database,Tumor Immune Estimate Resource(TIMER)database and Human Protein Atlas(HPA)database to evaluate SHC1 expression levels in normal and lung cancer tissues.Quantitative Real-time PCR(qRT-PCR)was performed in 15 pairs of NSCLC tissues and paired paracancer tissues.Kaplan-meier Plotter database and GEPIA2 database were used to analyze the relationship between SHC1 and prognosis in Lung adenocarcinoma(LUAD)and Lung squamous cell carcinoma(LUSC).Kaplan-meier Plotter database was used to analyze the correlation between SHC1 expression and prognosis in LUAD and LUSC with different clinicopathological factors,then hazard ratio(HR),95%confidence interval and Logrank P value were calculated.The Wanderer database was used to assess SHC1 methylation levels in LUAD and LUSC patients.To further investigate the relationship between the levels of each methylation probe and prognosis,we performed prognostic analysis using the MethSurv database.Results:The results of Oncomine,TIMER and HPA databases showed that the expression of SHC1 mRNA and protein was significantly increased in LUAD and LUSC tissues compared with normal tissues.Furthermore,we evaluated the expression of SHC1 mRNA in 15 pairs of NSCLC and paracancer tissues by qRT-PCR,and found that the expression of SHC1 mRNA in lung adenocarcinoma and lung squamous carcinoma were higher than that in paracancer tissues,which was consistent with the predicted results of the database.Kaplan-Meier Plotter database and GEPIA2 database analysis showed that patients with high SHC 1 expression in LUAD had poor prognosis,while patients with different SHC1 expression in LUSC had no significant difference in prognosis.Further subgroup analysis using the Kaplan-Meier Plotter database showed that high SHC1 expression was associated with poorer outcomes in women and patients with stage M0.In addition,high SHC1 expression was associated with worse prognosis in patients with stage Ⅰ and Ⅱ lung cancer,but no significant difference was seen in patients with stage Ⅲ lung cancer.The result of Wanderer database showed that methylation level of SHC1 was evaluated.In the LUAD dataset,probe level analysis revealed 10 obvious hypermethylated probes(CG07818949,CG00147095,CG09025625,CG24927174,CG06186450,CG04995846,CG22018051,and CG162)in the selected region of SHC1 in LUAD tissue 54756,CG19356022 and CG25603883);In CpG island,CG10892866,CG02341811,CG25151638,CG15556748,and CG02293828 also showed differences in methylation levels between LUAD and normal tissues.In the LUSC dataset,Probe level analysis revealed nine distinct hypermethylated probes(CG07818949,CG00147095,CG09025625,CG24927174,CG06186450,CG04995846,CG22018051,and CG1625)in selected SHC1 regions of LUSC tissue 4756 and cg02576073);On CpG Island,Cg10892866,CG14024356,CG27105205,CG02341811,CG24683561,CG18188585,CG01277844,CG25251638,CG15556748,CG00915289 and CG00993057 had significant differences in methylation levels between LUSC tissue and normal tissue.Survival analysis in MethSurv database showed that patients with high methylation levels of CG04995846,CG06186450,CG24927174,CG12473916,CG19356022 and CG25603883 in LUAD had a better prognosis.Patients with high methylation levels of CG07818949,CG12473916,and CG19356022 in LUSC had a better prognosis.Conclusions:This study is the first to explore the clinical significance of SHC1 and its methylation in LUAD and LUSC.SHC1 expression and cg12473916 and CG19356022 methylation levels may provide more accurate prognostic information for LUAD,but their prediction valve for LUSC may be limited.