A nomogram model for predicting distant metastasis of newly diagnosed colorectal cancer based on clinical features

Abstract

ObjectiveColorectal cancer is one of the most common primary malignancies and the third most common cause of cancer death in both men and women in the United States. Among people diagnosed with initial colorectal cancer, 22% had metastatic colorectal cancer, while the 5-year survival rate was less than 20%. The purpose of this study is to develop a nomogram for predicting distant metastasis in newly diagnosed colorectal cancer patients and to identify high-risk groups.MethodsWe retrospectively reviewed the data of patients who were diagnosed with colorectal cancer at Zhong nan Hospital of Wuhan University and People’s Hospital of Gansu Province between January 2016 and December 2021. Risk predictors for distant metastasis from colorectal patients were determined by the univariate and multivariate logistic regression analyses. Nomograms were developed to predict the probabilities of distant metastatic sites of colorectal cancer patients and evaluated by calibration curves, receiver operating characteristic curves, and decision curve analysis (DCA).ResultsA total of 327 cases were included in this study: 224 colorectal cancer patients from Zhong nan Hospital of Wuhan University were incorporated into the training set, and 103 colorectal cancer patients from Gansu Provincial People’s Hospital were incorporated into the testing set. By univariate logistic regression analysis, platelet (PLT) level (p = 0.009), carcinoembryonic antigen (CEA) level (p = 0.032), histological grade (p < 0.001), colorectal cancer tumor markers (p < 0.001), N stage (p < 0.001), and tumor site (p = 0.005) were associated with distant metastasis in colorectal cancer patients. Multivariate logistic regression analysis showed that N stage (p < 0.001), histological grade (p = 0.026), and colorectal cancer markers (p < 0.001) were independent predictors of distant metastasis in patients initially diagnosed with colorectal cancer. The above six risk factors were used to predict distant metastasis of newly diagnosed colorectal cancer. The C-indexes for the prediction of the nomogram were 0.902 (95% confidence interval (CI), 0.857–0.948).ConclusionThe nomogram showed excellent accuracy in predicting distant metastatic sites, and clinical utility may facilitate clinical decision-making.

A machine learning-based model for predicting distant metastasis in patients with rectal cancer

“…Chang et al. developed a model that incorporated a small sample size of data, making the developed model potentially biased ( 36 ). The paper uses big data from SEER to create the model, uses external data to validate the model, and finally develops a clickable web calculator to aid the clinical dissemination of the model.…”

“…Secondly, the method used to construct the models is logistic regression, which has specific requirements for data distribution and is sensitive to multivariate covariance and therefore has some limitations in its application (15). Chang et al developed a model that incorporated a small sample size of data, making the developed model potentially biased (36). The paper uses big data from SEER to create the model, uses external data to validate the model, and finally develops a clickable web calculator to aid the clinical dissemination of the model.…”

Predicting colorectal cancer risk: a novel approach using anemia and blood test markers

“…Machine learning techniques have significantly contributed to the evaluation of metastasis and prognosis in contemporary studies on colorectal cancer, exemplified by the utilization of the nomogram model (9), the 9-gene COX regression model (10), the random forest model (11), and the social ecological model (SEM) (12). These models employ a comprehensive approach to assess the pre-onset or post-onset condition of colorectal cancer in a population by simultaneously considering multiple risk factors.…”

Prediction of liver and lung metastases in patients with early‐onset colorectal cancer by nomograms based on heterogeneous and homogenous risk factors

“…However, patients younger than 50 years old are more likely to be uninsured and have lower compliance with screening, even if they have a family history of CRC. 5 Additionally, half of EOCRC patients are younger than 45 years old and therefore may not participate in screening. The increase in EOCRC has been accompanied by a decrease in late‐onset CRC, with the median age at diagnosis dropping from 72 years in the early 21st century to 66 years currently.…”

“…The US Preventive Services Task Force and the American Cancer Society have recommended lowering the screening initiation age to 45 years. However, patients younger than 50 years old are more likely to be uninsured and have lower compliance with screening, even if they have a family history of CRC 5 . Additionally, half of EOCRC patients are younger than 45 years old and therefore may not participate in screening.…”