Nomogram Model Could Predict Survival in Stage III CRC

According to International Journal of Colorectal Disease,¹ nomograms and predictive models were developed that could predict survival in stage III colorectal cancer (CRC) after operation. The model, which focused on patients with T3 and T4 CRC, maintained high stability after adjustment.

Lymph node metastases and T3 and T4 stage tumors are characteristic of colon cancer, with stage III CRC having a proportion of T3 or T4 stage tumors that can range between 84.3% and 91.6%. The prognosis for these tumors is quite low due to their invasive nature. Postoperative mortality and recurrence have been found to vary after surgery, with mortality changing directly following operation through follow-up. Postoperative duration and tumor size can both affect patient prognosis.

This study aimed to evaluate the risk factors that are associated with conditional survival rates, including conditional overall survival (cOS) and conditional cancer-specific survival (cCSS), in patients who had surgery for stage III T3-T4 colon cancer. Predicting the probability of survival through a new nomogram model was also a study objective.

All patients in this study had confirmed stage III T3-T4 colon cancer and had data extracted from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Cancer database. All data were from 2010 to 2019. Patients were included if their diagnosis of stage III colon cancer came between 2010 and 2019 and they had follow-up data. Patients with nonprimary tumors, invalid follow-up data, unclear tumor locations or pathological diagnoses, an unknown number of lymph nodes harvested, or an unclear tumor grade were excluded. Data on tumor stage, site, and size were collected along with number of lymph nodes harvested and the scope of the lymph nodes.

All patients were split into training (n = 23,186) and validation (n = 9931) groups for postsurgical outcomes. OS and CSS rates were 60.2% and 69.5%, respectively, for a 5-year period. Achieving a 5-year OS increased to 67.9% after 1 year, 75.3% after 2 years, 83.6% after 3 years, and 91.4% after 4 years. CSS had a similar trend, with an increase to 75.5% after 1 year, 82.0% after 2 years, 88.6% after 3 years, and 94.4% after 4 years.

Previous studies show that nomogram models are able to predict CRC using body mass index and age.² The present study instead used independent risk factors to create the nomograms that would predict postoperative cOS and cCSS through 1 year after surgery. Risks of cOS and cCSS was determined through drawing a line from the corresponding risk score to the total score. The postoperative cOS in the training cohort had C-index of 0.701 (95% CI, 0.711-0.691) and cCSS, 0.701 (95% CI, 0.713-0.689) after 1 year. cOS and cCSS had identical area under the curve values of 0.732, 0.728, 0.734, and 0.737 after 2, 3, 4, and 5 years, respecftively.

There were some limitations to this study. Survival bias is possible due to no external validation, and detailed information on treatment modalities are not included in the SEER database, which prevents assessment of all treatments in relation to survival.

The researchers concluded that the nomogram was able to accurately predict survival in patients who had stage III T3-T4 colon cancer. Due to possible survival bias, future studies should focus on confirming the results externally.

References

1. Zeng H, Xue X, Chen D, et al. Conditional survival analysis and real-time prognosis prediction in stage III T3-T4 cancer patients after surgical resection: a SEER database analysis. Int J Colorectal Dis. 2024;39(1):54. doi:10.1007/s00384-024-04614-x
2. Bonavitacola J. Early screening model displays high accuracy in predicting CRC. The American Journal of Managed Care^®. March 7, 2024. Accessed April 23, 2024. https://www.ajmc.com/view/early-screening-model-displays-high-accuracy-in-predicting-crc