This study revealed that the presence of TDs is an independent, unfavorable prognostic factor for patients with sigmoid colon cancer (hazard ratio [HR] 1.310, 95% CI 1.049–1.638, p = 0.017), and that their presence/absence could guide decisions regarding treatment strategies. Not all stage N1c patients should receive chemotherapy, which is recommended for positive LN patients. Patients aged < 70 years, those of black race, and those with stage T1–2 showed similar curative prognoses when receiving surgical treatments alone, as when receiving a combination of surgery and chemotherapy (all p > 0.05). Thus, patients without other high-risk factors may be able to forgo chemotherapy in clinical practice. Using the results of multivariable analysis and VIMP values obtained through RF, factors that have a significant impact on OS were selected to establish a nomogram model for predicting one-, three-, and five-year OS rates. The new model had better discrimination than the AJCC’s TNM model (C-index: 0.728 versus 0.667, respectively, p < 0.001; IDI = 0.074, p < 0.001), as well as good consistency and higher clinical benefit.
The National Comprehensive Cancer Network’s guidelines suggest that TDs are related to the N stage. However, some studies have suggested that patients with stage N1c can be incorporated into stage N1b or stage N2a groups, in terms of their survival20,21. Some studies also show that the N stage can be further subdivided by counting TDs as positive LNs7,22. Our study demonstrated that patients with stage N1c have worse OS than patients with stage N0 (p < 0.001). Further, our multivariable analysis revealed that patients with stage N1c or stage N1a have a similar risk of death as those with stage N0 (HR 1.310, 95% CI 1.049–1.638; HR 1.345, 95% CI 1.237–1.463). These results indicate that the role of TDs in sigmoid colon cancer is complicated and that they may play important roles in tumor proliferation and invasion.
Nevertheless, the exact effect and mechanism of TDs in prognosis remain unclear. Ueno find that the most prominent feature of TDs is undifferentiated cancer cell microscopic clusters in adipose tissue, which is similar to tumor budding at the invasive front of the main tumor23. Numerous studies have shown that tumor budding may be the early stage of invasion and related to LN infiltration and poor prognosis24,25. Therefore, TDs may also indicate that a tumor is more invasive. In addition, Prabhudesai find a marked association between TDs and vascular invasion, suggesting that TDs represent bloodborne spread and an early indicator of distant metastasis of tumor cells26. Therefore, it is reasonable to conclude that patients with TDs will have a worse prognosis.
We obtained new insights regarding therapeutic strategies for patients with stage N1c disease without distant metastasis. Surgery combined with adjuvant chemotherapy is recommended for patients with stage III colon cancer. Bouquot find that stage N1c is associated with a higher T stage, and suggests that such patients be considered high-risk and candidates for adjuvant chemotherapy27. However, the present study found that not all patients with stage N1c sigmoid colon cancer require chemotherapy following surgery. For patients younger than 70 years, those with stage T1–2 and those of black race receiving chemotherapy did not impact prognosis (all p > 0.05). If patients with these characteristics do not have high-risk factors—such as positive resection margins, neurovascular invasion, or vascular tumor thrombus—surgery alone, and enhanced follow-up monitoring, could be considered to thereby avoid subjecting patients to unnecessary treatment. Previous studies had shown that older adults have poor resistance to chemotherapy and cannot benefit from it. Nevertheless, it has been shown that older adult patients (≥ 75 years old) with stage III colon cancer can benefit from adjuvant chemotherapy compared to those who did not receive adjuvant chemotherapy (3-year overall survival: 60–71% versus 50–53%)28. The International Society of Geriatric Oncology (SIOG) also believed that older adult patients with stage III colon cancer can obtain the same benefits as young people from adjuvant chemotherapy29,30. Therefore, for N1c stage III older adult patients, after evaluating chemotherapy tolerance, adjuvant chemotherapy may be considered to obtain better survival rates.
For patients with stage N0, receiving surgery combined with chemotherapy showed no impact on OS, compared to receiving surgery alone (p = 0.456). This is similar to the findings of other studies. The British Quick and Simple and Reliable trial (commonly referred to as the QUASAR trial) and a previous meta-analysis conclude that the 5-year OS benefit of fluorouracil-based adjuvant chemotherapy is only 3–4% (95% CI 1.0–6.0), suggesting that it fails to significantly improve the prognosis of patients with stage II colon cancer31,32. Further, the MOSAIC trial indicated that adjuvant oxaliplatin-based chemotherapy does not provide survival benefits for patients with stage II colon cancer, unless they have high-risk clinical features33.
For both stage N0 and N1c patients, our study did not find a prognostic benefit of radiotherapy. Currently, the available data on whether radiotherapy can increase OS rates in patients with colon cancer is inconsistent34. Retrospective studies from the 1980s and 1990s have demonstrated improvements in local control (LC) and disease-free survival (DFS) after adjuvant external beam radiotherapy35,36. However, in a subsequent randomized controlled trial, no differences in DFS or OS were observed between the test groups, which was inconsistent with the previous conclusions37. Meanwhile, a recent retrospective study reports that adjuvant radiotherapy can improve LC and DFS in some patients with colon cancer, such as those with stage T4b and/or residual tumors38. However, organ movement is an important factor affecting the accuracy of radiotherapy. Some studies indicate that respiratory movement seriously affects the effect of radiotherapy on CRC tumors, which increases the difficulty of using radiotherapy to treat colon cancer39,40. Currently, the clinical application of radiotherapy in colon cancer is insufficient, meaning there is a lack of data concerning exposure ranges, radiation dosages, and fraction numbers; this lack of data contributes to the limited application of radiotherapy, forming a vicious circle. Therefore, the role of radiotherapy in colon cancer requires further investigation.
As sigmoid colon cancer has a poor prognosis, it is important to determine prognostic factors. Previous researches find that age, tumor differentiation, T stage, and N stage are independent factors influencing survival, of which T stage is the most significant factor41,42. The present study also confirmed the role of these factors in prognosis, using multivariable analysis. Meanwhile, we also found sex, race, marital status, serum CEA level, tumor size, PNI, and LND number to be independent predictors of OS (all p < 0.05), suggesting that there are numerous factors affecting the survival of patients with sigmoid colon cancer. Therefore, it is necessary to determine key prognostic characteristics. We calculated and ranked the influence of each factor on survival through the machine learning method RF and found that the effects of age, T stage, and N stage on survival are greater than those of other factors (VIMP = 0.0724, 0.0387, and 0.0285, respectively). Benefiting from the high stability of RF-calculated results, the established prediction model had good generalization ability.
Furthermore, we constructed a nomogram based on age, T stage, and N stage to predict the 1-, 3-, and 5-year OS of patients with sigmoid colon cancer. To our knowledge, this is the first predictive model for sigmoid colon cancer survival. The nomogram showed that the scores for the different age, T stage, and N stage groups had large ranges, suggesting that these factors have a greater impact on prognosis. Among these, age showed the greatest influence on OS; this may be related to the limited remaining life of older adults, as well as their poor tolerance of extensive surgery, radiotherapy, and chemotherapy. As an example for explaining the application of the model, consider a patient with sigmoid colon cancer who is aged 61 years (43 points), and is postoperative stage T3 (58 points), N2a (75 points), and M0. For this patient, the total risk value would be 176, trending downward on the “1-, 3-, and 5-year overall survival” axes. The 1-year survival rate would be 93.0%, the 3-year survival rate would be 79.0%, and the 5-year survival rate would be 68.0%.
In addition, the nomogram demonstrated better discrimination in predicting OS, compared to the TNM staging system, showing a C-index of 0.728 in comparison to the TNM’s C-index of 0.667 (p < 0.001). IDI was 0.074 (p < 0.001), indicating that the predictive capability of the new model was improved by 7.40%. For internal validation, bootstrap resampling with 1000 iterations was performed, and the calibration plots exhibited good consistency between the predicted and actual OS. Finally, the DCA for the first, third, and fifth years showed that, in clinical practice, predicting the survival of patients using this novel nomogram can bring higher benefits for patients than using the TNM staging model. This may be because different expected survival times can cause different psychological burdens for patients, and can also have an impact on subsequent treatment and follow-up strategies.
There are some limitations to this study. First, the existence of missing data in samples inevitably introduced a potential selection bias and accuracy loss. Second, because of the limitations of the SEER database, some information was not included, including tumor recurrence, details on chemotherapy regimens, and radiation plans. Third, the SEER database lacks information such as surgical margin status, microsatellite status, and Ki-67 proliferation index; therefore, these variables are not included in the prediction model. Fourth, it is important to highlight that the nomogram created in this study only applies to patients without distant metastasis and who had undergone surgery. Lastly, this was a retrospective study, and further prospective, multicenter, large-scale studies are needed to verify our findings.