You are here
Home > Blog > Gastroenterology > Colorectal Cancer: A Nomogram To Predict Metastases

Colorectal Cancer: A Nomogram To Predict Metastases

Colorectal Cancer: A Nomogram To Predict Metastases


This study addresses the imperative task of identifying risk factors associated with distant metastasis in early-onset colorectal cancer (EOCRC), aiming to enhance understanding of its etiology and facilitate preventive treatments. The research, utilizing data from EOCRC patients in the SEER database (2010-2015), seeks to characterize the variability in EOCRC incidence and discern both heterogeneous and homogeneous risk factors linked to synchronous liver, lung, and hepato-lung metastases.

The methodology involves categorizing patients into three groups based on synchronous liver, lung, and hepato-lung metastases, with random assignment to development and validation cohorts. Logistic regression is employed to analyze risk factors, and nomograms are constructed for risk calculation. Evaluation metrics such as ROC curves and calibration curves quantitatively assess the predictive performance.

With a total of 16,336 eligible EOCRC patients, 17.90% experienced distant metastases. The incidences of synchronous liver, lung, and hepato-lung metastases were 11.90%, 2.42%, and 1.50%, respectively. Positive CEA values pre-treatment, increased lymphatic metastases, and deeper invasion of the intestinal wall exhibited positive correlations with all three distant metastases. Conversely, the correlation of age, ethnicity, primary tumor location, and histologic grade among the three types was inconsistent. The ROC curve and calibration curve demonstrated robust performance in predicting distant metastases in EOCRC.

In conclusion, this study reveals significant differences in the incidence of distant metastases in EOCRC, with risk factors exhibiting both heterogeneous and homogenous associations. Despite incorporating limited risk factors, the established nomograms demonstrate commendable predictive performance.


Colorectal cancer (CRC) is a significant global health concern, ranking as the third most prevalent cancer and the second leading cause of cancer-related deaths worldwide. Over recent decades, there has been a notable rise in early-onset colorectal cancer (EOCRC), particularly among those under 50 years old. This contrasts with a decline in late-onset CRC cases due to effective screening, surveillance, and treatment. EOCRC’s incidence has surged by 45% in the past 30 years, prompting recommendations to lower screening initiation age to 45, yet challenges like lower compliance persist, especially in the uninsured and those under 50.

Tumor metastasis is a primary contributor to CRC-related fatalities, emphasizing the need for early detection and intervention. EOCRC patients often face delayed diagnosis and lack awareness of symptoms, leading to advanced-stage diagnoses and increased mortality. Liver and lung are common metastatic sites, influencing prognosis. Imaging technologies play a crucial role in detecting metastasis, with CT as the primary modality, but late detections still occur.

This study aims to fill a research gap by investigating risk factors and prognostic variables specific to EOCRC patients, focusing on synchronous liver, lung, and hepato-lung metastases. A large cohort study, utilizing the SEER database, seeks to characterize incidence differences and develop nomograms for predicting risks. The goal is to aid clinicians in devising tailored medical strategies and treatments for EOCRC patients based on early identification of metastasis risk factors.


This population-based research utilized data from the open-access Surveillance, Epidemiology, and End Results (SEER) database, administered by the National Cancer Institute (NCI). The study focused on patients diagnosed with early-onset colorectal cancer (EOCRC) between 2010 and 2015, specifically those with synchronous liver, lung, and hepato-lung metastases at the time of initial colorectal cancer diagnosis.

Inclusion criteria involved excluding cases diagnosed at autopsy or via death certificates, those aged <18 or ≥50, cases pathologically diagnosed as in situ carcinoma or lacking a pathological diagnosis, and instances with an unknown primary tumor location. Patients who did not have information on distant metastases were not included in the study. The final study sample comprised three groups: liver metastases (N =16,146), lung metastases (N =16,126), and hepato-lung metastases (N =16,108). Each group was randomly divided into development and validation cohorts in a 7:3 ratio, with specific allocations for liver, lung, and hepato-lung metastases groups. The development cohort determined independent risk factors and model construction, while the validation cohort served for internal model validation. The SEER*Stat version 8.4.1 software was employed to generate the case list.

Statistical Analysis

The presentation of data in this study included categorical data expressed as numbers and percentages (N, %), while quantitative data were presented as means±standard deviations (SD). Statistical comparisons of categorical variables employed Chi-square tests. Univariate and multivariate logistic regression models were utilized to identify risk factors associated with distant metastasis in early-onset colorectal cancer (EOCRC). Factors exhibiting statistically significant differences in the univariate analysis were incorporated into the multivariate analysis.


The multivariate logistic analysis results were used to identify the intersection of independent risk factors for different types of metastases, assessing heterogeneity or homogeneity, and visualized using Venn diagrams. Predictive diagrams for synchronous liver, lung, and hepato-lung metastases in EOCRC were developed. Evaluation of the nomograms’ predictive efficacy involved calibration curves (with 1000 bootstrap samples), receiver operating characteristic (ROC) curves, and the calculation of the area under the curves (AUC). Statistical analyses were conducted using GraphPad Prism version 9.0 and SPSS version 26.0, with a significance level set at a two-tailed p-value less than 0.05. Venn diagrams were generated using Figdraw, and R version 4.2.1, incorporating the “rms” ​and “pROC” packages, was utilized for drawing the nomograms and ROC curves, respectively.


In this study, a total of 16,363 eligible patients diagnosed with early-onset colorectal cancer (EOCRC) between 2010 and 2015 were extracted from the SEER database. The final sample, after excluding cases with unknown information on distant metastases, comprised 16,336 patients. Among them, 1921 had liver metastasis only, 390 had lung metastases only, and 241 had both liver and lung metastasis. The average age was 41.72±6.79 years, with 51.23% males and 54.9% married. The majority were white, and the rectum was the most common site of EOCRC. The incidence of distant metastasis was 17.90%, with liver, lung, and hepato-lung metastases occurring in 11.90%, 2.42%, and 1.50% of cases, respectively.

Statistical analyses revealed significant associations between various demographic and clinical factors and the occurrence of distant metastasis. Multivariate logistic regression analysis identified several risk factors, including older age, right/left colon location, poor histological grade, mucinous adenocarcinoma, AJCC pT stage, lymph node metastasis, and positive CEA value before treatment.

Heterogeneity and homogeneity were observed in risk factors for specific organ metastasis. Nomograms were developed to predict the likelihood of synchronous liver, lung, and hepato-lung metastases, showing good concordance and predictive performance. The AUCs for liver, lung, and hepato-lung metastases were 81.8%, 80.2%, and 83.1% in the development cohort, and 77.3%, 71.9%, and 73.7% in the validation cohort, respectively. Optimal thresholds for sensitivity and specificity were also provided.

These findings contribute valuable insights into the risk factors and predictive models for synchronous distant metastases in EOCRC, providing clinicians with tools to enhance early identification and management strategies for high-risk individuals.


This study addresses the dearth of comprehensive research on simultaneous liver, lung, and hepato-lung metastases in early-onset colorectal cancer (EOCRC), representing the largest investigation in this domain. The liver emerges as the primary organ affected by metastasis in EOCRC. Significant clinical and pathological differences exist among liver, lung, and hepato-lung metastases, emphasizing the need to identify independent risk factors for organ-specific metastasis. The study reveals both heterogeneity and homogeneity in factors associated with distant metastasis at different sites in EOCRC.

Distinct risk factors are identified for liver, lung, and hepato-lung metastases, such as age, primary tumor location, histological grade, ethnicity, and other clinical parameters. Notably, risk factors for hepato-lung metastases differ from conventional patterns observed in colorectal cancer across all age groups. The study explores the heterogeneity of risk factors based on tumor location, demonstrating that the sigmoid colon is most prone to distant metastases in EOCRC, differing from established trends in older patients.

The findings suggest potential molecular and genetic distinctions contributing to metastasis in EOCRC, warranting further exploration. The study introduces prediction nomograms based on identified risk factors, offering a cost-effective and valuable tool for clinicians to assess high-risk EOCRC patients. However, limitations include the absence of certain risk factors in the dataset and the study’s focus on the American population, necessitating external validation for broader applicability. Despite these limitations, the nomograms exhibit favorable predictive performance.

In conclusion, this study enhances understanding of distant metastasis in EOCRC, offering insights into organ-specific risk factors. The nomograms present a promising method for initial screening, potentially improving early detection and personalized treatment strategies for this patient population.

Oncology Related Tools


Latest Research

Colorectal Cancer

About Author

Similar Articles

Leave a Reply