Prognostic analysis and beneficiary identification of adjuvant external beam radiotherapy for stage pT4b sigmoid colon cancer

The clinical efficacy of adjuvant radiotherapy in sigmoid colon cancer remains questioned. To evaluate the clinical efficacy of adjuvant external beam radiotherapy (EBRT) for patients with pathologic stage T4b sigmoid colon cancer. Patients with stage pT4b sigmoid colon cancer receiving adjuvant EBRT or not followed by surgery between 2004 and 2016 were extracted from the Surveillance, Epidemiology, and End Results database. Analysis of overall survival (OS) was performed using Kaplan–Meier curves and prognostic factors were identified using Cox proportional hazards regression models with 95% confidence intervals within the entire cohort. A risk-stratification system was then developed based on the β regression coefficient. Among 2073 patients, 284 (13.7%) underwent adjuvant EBRT. The median OS in the group receiving adjuvant EBRT was significantly longer than that in the non-radiotherapy group (p < 0.001). Age, serum carcinoembryonic antigen (CEA) level, perineural invasion, lymph node dissection (LND) number, and adjuvant EBRT were independent factors associated with OS. A risk‐stratification system was generated, which showed that low‐risk patients had a higher 5-year survival rate than high-risk patients (75.6% vs. 42.3%, p < 0.001). Adjuvant EBRT significantly prolonged the 5-year survival rate of high-risk patients (62.6% vs. 38.3%, p = 0.009) but showed no survival benefit among low‐risk patients (87.7% vs. 73.2%, p = 0.100). Our risk‐stratification model comprising age, serum CEA, perineural invasion, and LND number predicted the outcomes of patients with stage pT4b sigmoid colon cancer based on which subgroup of high-risk patients should receive adjuvant EBRT.


Methods
Study design and ethics statement. This retrospective study analyzed data from the publicly accessible SEER database. Before the study, we obtained an official permit for the research purpose (ID: 12284-Nov2019). Informed consent or ethical approval was not required for this study.
Patient selection and data extraction. Cases  were initially identified as eligible for this study. Among these, 3880 patients were excluded as follows: (1) 1297 patients had multiple primary tumors; (2) 1187 patients did not undergo surgery or local tumor excision; (3) 83 patients received preoperative radiotherapy or intraoperative radiotherapy or both preoperative and postoperative radiotherapy or an unknown sequence; and (4) 1313 patients were with M1 or unknown. All patients underwent active follow-up. Finally, 2073 patients, including 284 (13.7%) patients receiving radiotherapy and 1789 (86.3%) patients without radiotherapy, were further analyzed (Fig. 1). The T classification was restaged according to the 8th edition of the American Joint Committee on Cancer (AJCC) staging classification system based on the following codes: derived AJCC T, 6th (2004+); derived AJCC, 7th (2010+); and Collaborative Stage (CS) tumor size (2004+) 18 .

Statistical analyses.
The primary endpoint in this study was OS. Kaplan-Meier (K-M) survival curves were determined using log-rank tests. Univariate analysis was conducted on all variables in this study; those with p < 0.05 were included in the multivariate Cox regression models to estimate the potential predictors associated with OS.
An easy risk score model was established using the β regression coefficient to predict the prognosis of patients with stage pT4b sigmoid colon cancer, which was assessed by receiver operating characteristic (ROC) curve analysis 22 . The cutoff value affecting OS was determined using the Youden index 22 . K-M survival curves were then used to evaluate the role of adjuvant radiotherapy in the treatment of patients with stage pT4b sigmoid colon cancer stratified according to risk scores. www.nature.com/scientificreports/ Statistical tests were conducted using RStudio, including the Table 1, survminer, and survival packages, or IBM SPSS Statistics for Windows, version 24.0. A p value < 0.05 was considered statistically significant.
Ethical approval. For this type of study formal consent is not required.
Informed consent. As the data used was from SEER dataset (public). Consent to participate could be checked in SEER.

Results
Patient characteristics. The clinicopathological characteristics of the 2073 patients eligible for inclusion in this study are summarized in Table 1. The proportions of patients aged ≥ 70 years, tumor size ≥ 5 cm, and LND number ≥ 12 were 39.1%, 72.9%, and 77.0%, respectively; however, only 13.7% of patients received adjuvant EBRT.
Establishment of the risk-stratification model. A risk score model was established according to the β regression coefficient and Exp (B) derived from the Cox model (Table 3). Briefly, age ≥ 70 years was scored as three points, elevated serum CEA as two, PNI as two, and LND number < 12 as three.
After excluding patients with missing information for any of the four variables, only 672 (32.4%) patients remained for further analysis. These remaining patients were scored according to the new risk score model. The area under the ROC curve was 0.703, and the optimal cutoff value was 2.5 points (Fig. 3a). According to this cutoff value, 354 (52.7%) patients were classified as having a low risk of poor prognosis (total score < 2.5), while 318 (47.3%) patients had a high risk of poor prognosis (total score ≥ 2.5). The median OS was significantly longer in the low-risk group than in the high-risk group, with 5-year survival rates of 75.6% and 42.3%, respectively (p < 0.001, Fig. 3b).
Furthermore, there was no significant difference in the median OS between low-risk patients with and without adjuvant radiotherapy (p = 0.100, Fig. 4a). Nevertheless, the median OS in high-risk patients receiving adjuvant radiotherapy was significantly longer than that in patients who underwent surgery alone (not reached vs. 44 months, p = 0.009, Fig. 4b), with 5-year survival rates of 62.6% vs. 38.8% (Table 4).

Discussion and conclusions
To the best of our knowledge, this is the first study to establish an easily assessed risk-stratification model to identify a subgroup of patients with stage pT4b sigmoid colon cancer who might benefit from adjuvant EBRT. According to the new risk stratification model, patients with a total score ≥ 2.5 would benefit more from adjuvant EBRT (p = 0.009), whereas patients with a total score < 2.5 would not show a survival benefit from adjuvant EBRT (p = 0.100).
Adjuvant EBRT is not routinely used in the treatment of colon cancer. Retrospective studies in the 1980s and the 1990s originally reported improved local control (LC) and disease-free survival (DFS) by adjuvant EBRT 12,23,24 ; however, these findings were refuted by a subsequent randomized controlled trial (RCT, Intergroup-0130). No differences in DFS or OS were observed between arms of the trial; however, the results in this trial have always been challenging because of the high ineligibility rates and inclusion of T3 patients 7 . Considering that the patterns of treatment failure have changed in the era of systemic therapy, including targeted therapy and immune therapy 25 , the role of adjuvant EBRT has been re-recognized along with the exploration of preoperative neoadjuvant chemoradiation for colon cancer and decreased toxicity associated with radiation. In 2016, a retrospective single-institution study 5 reported that adjuvant EBRT could enhance LC and DFS (hazard ratio [HR] 0.044, p < 0.05; HR 0.145, p < 0.05, respectively) of patients with colon cancer, specifically those with T4b and/or residual tumors, findings that were confirmed using external data or national cancer database 8,26 .
Compared with other sites of colon cancer, adjuvant EBRT for sigmoid colon cancer has several advantages. First, the sigmoid colon is in a relatively fixed anatomical location, which facilitates the delineation of the gross target volume 16,17 . Second, the organs at risk of sigmoid colon cancer are generally fewer than those of other sites of colon cancer. Third, the radiation exposure dose limit for the colon is lower than that for the small intestine. In this study, among patients in the crude cohort, 284 (13.7%) patients with stage pT4b sigmoid colon cancer who received postoperative radiotherapy showed prolonged OS compared with that in patients who did not receive postoperative radiotherapy (p < 0.001). As is well known, one treatment size does not fit all. We successfully identified a subgroup of high-risk patients with stage pT4b sigmoid colon cancer who could benefit from adjuvant EBRT based on our newly developed risk-stratification model.
Whether age affects LC and DFS remains controversial 3,27,28 ; however, aging is an independent risk factor for poor prognosis of colon cancer, with different aging cutoff values 4,8,21,29 . In this study, these cutoff values were 50 and 70 years and the proportion of patients aged ≥ 70 years was as high as 39 CEA is a routine index used for colon cancer diagnosis and surveillance 30,31 ; however, it remains controversial whether elevated CEA levels are an independent risk factor for poor prognosis 14,19,32 . In this study, serum CEA was an independent risk factor for OS. Moreover, elevated serum CEA doubled the risk of death in patients with stage pT4b sigmoid colon compared with that in patients with normal serum CEA levels.
Lymph node evaluation is often a critical factor in predicting the prognosis of colon cancer and is also a deciding factor for postoperative treatment 2,33,34 . However, LND number other than lymph node classification was independently associated with OS, as confirmed in this study, although the cutoff values for LND number varied in previous reports 3,35,36 . In this study, the cutoff value of LND number was 12, as recommended by the National Comprehensive Cancer Network (NCCN) guideline; based on this threshold, LND number < 12 increased the risk of death by nearly two-fold.
PNI is an aggressive characteristic of colon cancer, with incidence rates of 15-32% 14,37 . In a retrospective study of 269 patients with colorectal cancer, Liebig et al. 38 reported that patients with PNI had a four-fold worse 5-year survival compared with that in patients without PNI, a finding confirmed by several other studies 20,35,39 . However, PNI was also reported to be not associated with poor prognosis in cecum adenocarcinoma or colon cancer 3,14 40 . In our study, PNI increased the risk of worse OS in patients with sigmoid colon cancer compared with the risk in those without PNI.
This study has some limitations. First, this was a retrospective analysis and selection bias was difficult to avoid. Second, data on local recurrence, a critical endpoint of adjuvant EBRT, were unavailable in the SEER database 41 . Third, while recommended by the NCCN guidelines, this study did not analyze adjuvant chemotherapy, which may have weakened our conclusions. Fourth, data on surgical margin status, microsatellite status, and Ki-67% were not obtained from the current SEER database 18,42 ; hence, the newly developed risk model did not include these variables. Finally, details on radiotherapy including clinical tumor volume, radiation technique, dose/fraction, and acute/late toxicity were not recorded 43,44 ; thus, the feasibility of adjuvant EBRT requires further study.
In conclusion, the results of this study confirmed the survival benefit of postoperative radiotherapy for pT4b sigmoid colon cancer. A risk-stratification model including the easily measured factors of age, serum CEA, PNI, and LND number was then successfully established. In this model, patients with pT4b sigmoid colon cancer with scores ≥ 2.5 were recommended to receive postoperative radiotherapy. www.nature.com/scientificreports/ www.nature.com/scientificreports/  www.nature.com/scientificreports/

Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.