Construction Of A Mathematical Model For Prognosis Of Stage Ⅱ/Ⅲ Colorectal Cancer Based On Biomarkers And Pathological Signatures

Objective: The influence of perioperative serum tumor marker levels on the prognosis of stage Ⅱ and ⅡI CRC was systematically investigated.By integrating the pathological indicators and clinical variables in the internal database,a mathematical model for predicting the prognosis of CRC was constructed and verified by the external database.Methods: Patients who enrolled in the study were pathological diagnosed stage Ⅱ and ⅡI CRC after radical surgery.Patients’ data were collected including perioperative serum tumor markers,other clinical variables and pathological indicates.Follow-up data mainly included DFS and OS.We analyzed the relationship between the level of preoperative tumor markers and the clinical variables,and explored the factors affecting the prognosis of CRC.We further constructed a mathematical model to predict the survival of CRC patients by integrating various factors affecting the prognosis of CRC,and verified it with an external database.K-M curves were used to show survival between groups.Univariate and multivariate analyses were did by Cox proportional hazards model.Survival differences between groups were compared by log-rank test.Results:1.The median follow-up was 45 months.735 enrolled patients were assessed based on the numbers of increased tumor markers.We found that these increased tumor markers were closely associated with clinical stage,T stage,N stage,tumor location,pathology type,differentiation,lymphatic invasion and vascular invasion(all p values < 0.05).Furthermore,the number of increased tumor markers directly affected the survival of patients with CRC after curative surgery.The 3-year DFS and OS of patients with a score of 0 were 84.0% and 91.0%,respectively,which are much higher than those of patients with a score of 4(42.9% and 37.8%,respectively)(p <0.05).The 5-year DFS and OS of patients with a score of 0 were 75.9% and77.9%,respectively,which are much higher than those of patients with a score of 4(31.7% and 23.6%,respectively).Interestingly,our results suggested that stage ⅡI CRC patients with a score of 0 had longer DFS and OS times than stage Ⅱ patients with scores of 3 and 4.Further analysis revealed statistically significant differences in OS(p < 0.05)but not in DFS.2.1008 patients from three centers who underwent radical surgery were enrolled.Our results showed that positive postoperative CEA and CEA increment were related to clinical stage,T stage,N stage,tumor differentiation and lymphatic invasion(p values < 0.05).Univariate and multivariable analyses results suggested that positive postoperative CEA and CEA increment were independent prognostic factors for PFS(HR=3.149,95% CI,2.426-4.088,p=0.000 for postoperative CEA;HR=2.708,95%CI,2.106-3.482,p=0.000 for CEA increment)and OS(HR=3.414,95%CI,2.549-4.574,p=0.000 for postoperative CEA;HR=2.373,95%CI,1.783-3.157,p=0.000 for CEA increment).The survival analyses revealed positive postoperative CEA and CEA increment predicted worse prognosis.Furthermore,our results indicated that the 3-year and 5-year PFS rates were 86.6% and 78.4% in group A,but decreased to 25.3%and 7.2% in group D(p < 0.001).Similarly,the 3-year and 5-year OS rates for group A were 92.5% and 83.9%,much higher than group D(p < 0.001).In other words,patients with both postoperative CEA elevation and CEA increment had the worst prognosis.3.In order to construct the CRC prognosis model,8 variables including T stage,N stage,tumor differentiation,lymphatic invasion,vascular invasion,preoperative CEA,CA125 and CA199 were included.Univariate and multivariate analyses showed that these indicators affected postoperative OS.The model constructed by further analysis C-index=0.72.All patients were divided into three groups: high risk,medium risk and low risk.External data verify that the accuracy and stability of the model are high.Conclusions:1.The number of increased tumor markers could significantly predict prognosis in stage Ⅱ and ⅡI CRC.In addition,these increased tumor markers had direct impacts on metastasis as well as the recurrence status and survival time of stage Ⅱ and ⅡI CRC patients.2.Positive postoperative CEA and CEA increment were independent prognostic factors for stage Ⅱ and ⅡI CRC.Additionally,postoperative CEA and CEA increment had significant impacts on PFS and OS of CRC.3.The mathematical model used to predict the survival of CRC by integrating pathological stages and clinical indicators has good accuracy and stability,which has high clinical application value.