Colon cancer is one of the most prevalent malignancies and the second leading cause of cancer-related death in the world (1, 2). According to the latest statistics, there are approximately 520,000 new cases and 250,000 deaths in 2018 in China (3). Currently, the main treatment for colon cancer is radical surgical resection and adjuvant chemotherapy. The 5-year overall survival (OS) rate after radical surgery for colon cancer is approximately 60% in China (4), and the mortality rate remains high. Various factors might be associated with 5-year survival, such as stage, differentiation, histology, tumor location, age, and sex.

Accurate pathological diagnosis, radical surgery, and adequate postoperative adjuvant chemotherapy are critical for colon cancer patients. There have been extensive studies evaluating how to predict the postoperative OS of colon cancer; however, the results are controversial. Therefore, we retrospectively reviewed the clinicopathological data of 2,655 colon cancer patients after radical surgery between January 2011 and December 2016 at Harbin Medical University Cancer Hospital. A nomogram was developed based on clinical and histopathological high-risk factors to determine the OS of colon cancer patients undergoing curative resection.

The clinical data of 2,655 colon cancer patients treated between January 2011 and December 2016 were derived from the Department of Colorectal Surgery of Harbin Medical University Cancer Hospital. Patients who received palliative surgery, had multiple primary cancers, or received neoadjuvant chemotherapy were excluded; patients lost to follow-up were also excluded ( Supplementary Figure 1 ). All patients were randomly divided into a training cohort and a testing cohort at a ratio of 70% to 30%. This study was approved by the Harbin Medical University Cancer Hospital ethics committee (KY2017-19), and written informed consent was obtained from the patients.

We evaluated the following baseline covariates: sex, age, pathological tumor (pT) stage, pathological lymph node (pN) stage, pathological metastasis (pM) stage, tumor location, tumor size, histology, differentiation, examined lymph nodes (LNs), metastatic lymph node ratio (LNR), perineural invasion, lymphatic invasion, preoperative carcinoembryonic antigen (CEA) level, preoperative carbohydrate antigen 199 (CA199) concentration, and adjuvant chemotherapy. Cancers of the proximal splenic flexure (cecum and ascending colon) were classified as right colon cancer. Cancers in the distal splenic flexure (descending colon and sigmoid colon) were classified as left colon cancer. Cancer in the middle of the transverse colon was classified as transverse cancer. Regarding adjuvant therapy, patients were classified into two groups (i.e., with vs. without). Patients were also categorized into three groups by their age at the time of the primary tumor resection, as follows: <40 years, 40–70 years, and >70 years. CEA level was classified as ≤5 ng/ml or >5 ng/ml based on the upper normal limit (5 u/ml), and CA199 level was classified as ≤37 U/ml or >37 U/ml based on the upper normal limit (37 U/ml). Tumor size was categorized as <5 cm or ≥5 cm. All patients in this study were reclassified in accordance with the 8th TNM classification of colon cancer.

All statistical analyses were performed using SPSS software (version 25.0; IBM Corp., Armonk, NY, USA). OS analyses were performed using the Kaplan–Meier (KM) method, and the results were compared using the log-rank test. Clinical and pathological variables associated with survival were assessed on the basis of clinical importance, scientific evidence, and predictors identified in previously published articles (5, 6). The variables screened by univariate Cox regression, best subset regression (BSR), and least absolute shrinkage and selection operator (LASSO) were included in a multivariate Cox regression model, and the final model of the three methods was determined using backward stepwise selection with minimum Akaike’s information criterion (AIC) values ( Supplementary Table 1 ). Candidate variables with p-values < 0.2 in the univariate analysis were included in the multivariable model. Risk factors were measured as [hazard ratio (HR), 95% confidence interval (CI)]. All reported statistical significance levels were two-sided, with statistical significance set at 0.05. The predictive nomogram was developed based on a multivariate analysis using the rms package in R, version 4.1.2 (www.r-project.org). Calibration plots, Harrell’s concordance index (C-index), area under the curve (AUC) values, and decision curve analysis (DCA) were used to evaluate calibrating ability.

In our study, the OS rates for stage I, II, III, and IV colon cancer were 96.6%, 88.7%, 69.9%, and 34.3%, respectively. Stage I and II survival rates were significantly higher than those for stage III and IV, which was consistent with previous studies (11). This difference might be related to tumor invasion and cell infiltration (12–14). To improve the survival of colon cancer patients, it is necessary to focus on prevention and early detection and treatment.

The key prognostic factors for colon cancer, especially the high-risk factor of stage II disease, have been a hot topic of research. The poor prognosis for stage II colon cancer has been confirmed to be associated with high-risk factors (15, 16). In our study, the presence of a higher number of high-risk factors was confirmed to be associated with worse OS based on the risk factor grouping. Additionally, adjuvant chemotherapy improved the survival rate in stage II patients with high-risk factors in our study. This is consistent with the results of earlier studies, including the QUASAR trial (17) and some retrospective analyses (18, 19). The recommendation that patients with stage II colon cancer with high-risk factors should receive postoperative adjuvant chemotherapy is also consistent with the current guidelines (20). Previous studies have shown that stage II patients with dMMR have a better prognosis than stage II patients with pMMR. However, another study indicated that pMMR should not be the only risk factor considered when determining whether to administer adjuvant chemotherapy in patients with stage II colon cancer or not (21).

The prognostic role of CEA and CA199 in predicting survival in colon cancer patients has been frequently studied; however, there is currently no reliable evidence confirming this role (22). Our study found that preoperative CEA and CA199 levels were independent prognostic factors in colon cancer. Elevated preoperative CEA and CA19-9 levels were significantly associated with worse OS. Furthermore, patients with both elevated CEA and CA199 levels had the worst 5-year OS rate. The same results were seen in stage II colon cancer patients. In a recent study, researchers found that CEA and CA199 were independent predictors of colon cancer recurrence and survival (23). However, there is no randomized controlled trial to evaluate whether adjuvant chemotherapy should be recommended for stage II patients without high-risk factors and with both elevated preoperative CEA and CA199. Our findings suggested that high preoperative levels of CEA and CA199 significantly affected the prognosis and survival of stage II colon cancer patients and that these factors can be high-risk factors for the need for chemotherapy in stage II colon cancer. This finding agrees with the results of a retrospective analysis by Thirunavukarasu et al. (24). There are also other clinical or pathological factors that significantly affect prognosis and survival in stage II patients.

Another issue that we evaluated in this study was the number of LNs examined. Pathological N staging reflects a strong association between the number of positive LNs and survival (6, 25). Notably, in this study, the 5-year OS rate for stage II patients with an insufficient number of harvested LNs (< 12) was 84.6%, while the 5-year OS rate for patients with a sufficient number of harvested LNs was 89.7%. However, there was no similar trend regarding LNs in stage III patients. This means that an insufficient number of LNs were examined, resulting in undetected positive LNs, especially in patients with stage II colon cancer. To avoid missing positive LNs, the NCCN guidelines recommend that the number of LNs collected should be more than 12 (9), which was further validated in our study.

A growing number of studies have shown that LNR is associated with prognosis, and that this ratio can be used as an independent prognostic tool to assess the prognosis of colon cancer patients (26, 27). However, currently, the optimal cutoff value for LNR is controversial. In our previous study, an information gain method was developed to redefine the cutoff value for LNR. The thresholds were 0, 0–0.3, 0.3–0.7, and >0.7. The study was based on a large sample of the Surveillance, Epidemiology and End Results database (SEER) to redefine the cutoff value for LNR (10). The thresholds of LNR were proved to be better than the previously reported studies (28, 29). Based on thresholds of 0, 0.3, and 0.7, the respective 5-year OS rates were 89.2%, 71.5%, 55.8%, and 34.5% in our study. There were different pN stages in each LNR group. In the LNR (0.3–0.7) group, we found that patients with pN1 stage colon cancer had worse 5-year OS than patients with pN2 stage cancer. Further observations showed that less than 12 LNs examined led to an underestimation of the pN stage. Moreover, the predictive effect of LNR was independent of the number of LNs examined, and other studies have found similar results (30, 31).

In this study, we developed and validated a nomogram for the postoperative individualized prediction of survival in colon cancer patients. The nomogram incorporates age, CEA, CA199, histology, pT stage, pM stage, LNR, perineural invasion, and adjuvant chemotherapy. Integrating the clinical and pathological risk factors into a convenient nomogram helps predict postoperative survival. It is important to perform an accurate risk stratification for patients with cancers such as colon cancer because the prognosis may be heterogeneous (32, 33). Nomograms may afford a more individualized method to provide prognostic information for patients compared with the TNM staging system. Compared with the TNM staging system, our nomogram had better performance for risk stratification of the prognosis of patients undergoing radical colon cancer surgery, with better AUC values and C-index.

TNM staging is indispensable for postoperative survival assessment of colon cancer and for determining whether postoperative adjuvant therapy is needed. Our results showed that elevated preoperative CEA and CA199 levels are poor prognostic factors in patients with colon cancer. Preoperative CEA and CA199 elevations should be considered as risk factors in stage II colon cancer patients. We propose that stage II colon cancer patients without high-risk factors and with both ele-vated preoperative CEA and CA199 should receive adjuvant therapy. Future randomized controlled trials to evaluate the role of adjuvant chemotherapy in these patients are needed. To avoid missing positive LNs and subsequent inadequate staging of LNs metastases, it is recommended that at least 12 lymph nodes be examined. LNR can be used as a supplement to pN staging when the number of LNs examined is insufficient. The nomogram described in this study predicted the survival of colon cancer patients better than that when using the TNM system.

Our study was a retrospective study performed in a single center. Although the study involved a large data sample, a study of colon cancer diagnosis, surgery, and postoperative treatment from a single center has limitations that may lead to selection bias. Hopefully, more data from multiple centers and larger samples will lead to more convincing conclusions.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Harbin Medical University Cancer Hospital ethics committee (KY2017-19). The patients/participants provided their written informed consent to participate in this study.

HL and JL contributed to the study conception and design. XP, BX, RW, and XW collected the data. XP and BX analyzed the data. XP, BX, JS, and ST interpreted the data. XP and BX drafted the manuscript. HL and JL contributed to critical revision of the manuscript. All authors read and approved the final manuscript.

This work was supported by the National Natural Science Foundation of China (grant nos. U20A20376 and 61972116) Beijing Award Foundation (grant nos. YXJL-2020-0818-0478) Wu Jieping Medical Foundation (grant no. 320.6750.2020-19-20) Heilongjiang Province Postdoctoral Science Foundation(grant no. LBHZ21189) Harbin Medical University Innovative Science Research Funded Project (grant no. 31041220028) China Postdoctoral Science Foundation(grant no. 2022MD713747).

We thank Jane Charbonneau, DVM, from Liwen Bianji (Edanz) (www.liwenbianji.cn) for editing the English text of a draft of this manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.