The Correlation Between Survival Benefit of Preoperative Radiotherapy and Pretreatment Carcinoembryonic Antigen Level in Locally Advanced Rectal Cancer

Background Preoperative radiotherapy followed by radical surgery is the standard treatment for locally advanced rectal cancer; however, its long-term survival benefit remains controversial. This study aimed to determine the relationship between pretreatment carcinoembryonic antigen (CEA) levels and the long-term prognosis of preoperative radiotherapy in locally advanced rectal cancer (LARC) patients. Methods Data of LARC patients who underwent surgery between 2011 and 2015 were identified from the Surveillance, Epidemiology, and End Results (SEER) database, and patients were accordingly divided into surgery (S) group and radiotherapy followed by surgery (RT+S) group. The primary outcomes were cancer-specific survival (CSS) and cancer-specific mortality (CSM). CSS was evaluated using Kaplan-Meier analysis, while CSM was evaluated using a competitive risk model. Subgroup analysis was also conducted, which was stratified by pretreatment CEA levels. Results A total of 2,760 patients were eligible for this study, including 350 (12.7%) patients in the S group and 2,410 (87.3%) in the RT+S group. There were no significant differences in the CSS and CSM rates at 1, 3, and 4 years between the S and RT+S groups before and after PSM (all p > 0.05). Pretreatment CEA levels were independently associated with CSS and CSM after adjusting for age, sex, stage, pathological factors, and treatment factors (all p < 0.05). Subgroup analysis showed that preoperative radiotherapy would benefit patients with elevated CEA in terms of CSS and CSM (both p < 0.05) but not those patients with normal CEA (both p > 0.05). Further analysis showed that preoperative radiotherapy was an independent protective factor for CSS and CSM in patients with elevated CEA levels (both p < 0.05). Conclusions Pretreatment CEA level may be considered a potential biomarker to screen LACR patients who would benefit from preoperative radiotherapy in terms of long-term prognosis.


INTRODUCTION
Preoperative radiotherapy, either long-course radiotherapy (LCRT) or short-course radiotherapy (SCRT), is the standard neoadjuvant strategy for locally advanced rectal cancer (LARC) (1,2). In the recent decade, the proportion of LARC patients receiving preoperative radiotherapy has been as high as 59.8% in the USA (3), although direct surgery is still preferred in some areas, such as Japan (4). With the advent of preoperative radiotherapy, the rate of sphincter preservation is increasing, this being mainly due to the significant downstaging effect (5,6). However, as a hard endpoint of treatment, the long-term survival benefit of preoperative radiotherapy remains controversial, regardless of LCRT and SCRT (7)(8)(9)(10)(11).
Carcinoembryonic antigen (CEA) is a routine screening and diagnostic index of colorectal cancer and is a widely used screening marker for postoperative recurrence (12)(13)(14)(15). CEA levels, both before and after surgery, have been identified as important risk factors for long-term prognosis as well as for dynamic changes in CEA levels (16,17). Moreover, CEA levels have been found to be associated with the response rate of neoadjuvant treatment (12,18). However, there are no reports on the use of CEA in guiding the management of preoperative radiotherapy in terms of long-term prognosis. In the current study, we selected LARC patients diagnosed between 2011 and 2015 in the Surveillance, Epidemiology, and End Results (SEER) database to identify the long-term survival benefit of preoperative radiotherapy and to determine the relationship between the pretreatment CEA level and the prognosis of patients receiving preoperative radiotherapy.

Ethics Statement
Since we gained an official permit to access the research data of the SEER database (ID: 22032-Nov2019) and all the analyses in the current study were conducted under the rules of the SEER database, neither informed consent nor ethical approval was required for this study.

Patient Selection
Patients were eligible if they: (1) underwent radical surgery and were diagnosed with rectal cancer by pathology, (2) staged at T 3-4 N any M 0 or T any N + M 0 , and (3) received chemotherapy. Patients were excluded from this study if they met one of the following criteria: (1) receipt of postoperative radiotherapy, (2) multiple cancers, (3) survival month ≤1 month, or (4) unknown clinical data. Based on whether patients received preoperative radiotherapy or not, they were divided into surgery (S) and radiotherapy +surgery (RT+S) groups.

Outcome Definition
The endpoints of this study were cancer-specific survival (CSS) and cancer-specific mortality (CSM). CSS was defined as the time from the date of diagnosis to the date of death from rectal cancer or the latest follow-up. CSM was defined as cumulative mortality from the date of diagnosis to the date of death from rectal cancer or at the latest follow-up.

Propensity Score Matching
Propensity score matching (PSM) analysis was performed to reduce selection bias. Briefly, baseline characteristics between the two groups were matched using the 1:1 nearest-neighbor matching method with a standard deviation of 0.2.

Statistical Analyses
The chi-square (c 2 ) test or Fisher's test was used for comparisons between the two groups. The Kaplan-Meier (K-M) method was used for comparison of CSS analysis between the two groups using a log-rank test. A multivariate Cox regression model was used to identify the independent risk factors for CSS.
In the competitive-risk analysis, death from other causes was recognized as a competitive event of cancer-specific death. Gray's test was used to determine the intergroup difference in the CSM, and the subdistribution proportional hazards model was used to perform multivariate analysis of CSM.
All statistical tests were conducted using RStudio (version 1.3.1073) in this study, including packages of xlsx, Table 1, survival, Survminer, MatchIt, cmprsk, and plyr. All tests were two-sided, and statistical significance was set at p < 0.05.

Patients' Characteristics
A total of 2,760 patients were eligible for this study, including 350 (12.7%) patients in the S group and 2,410 (87.3%) in the RT+S group. The baseline characteristics between the S and RT+S groups were unparalleled, as depicted in Table 1. Briefly, the proportions of males, elevated CEA level, LND <12, and T3 in the RT+S group were all higher than those in the S group (all p < 0.05, Table 1), while the rates of tumor differentiation grades III/IV, TD, PNI, stage III, and N1/2 were lower in the RT+S group than in the S group (all p < 0.05, Table 1). However, the baseline characteristics between the two groups were comparable after 1:1 PSM (all p > 0.05, Table 1).

Effect of Pretreatment CEA Level on CSS and CSM in LARC Patients
In the matched cohort, time-dependent coefficient analysis showed a strong association between CEA and CSS  Table 2). Furthermore, these associations were not attenuated after adjustment for all five factor groups (model 6). Likewise, pretreatment CEA level was also found to be an independent risk factor of CSM regardless of the model (all p < 0.05, Table 2).

Effect of Preoperative Radiotherapy on CSS and CSM in Normal CEA Subgroup
In the matched cohort, 420 patients had normal CEA levels, including 213 patients in the S group and 207 in the RT+S group. Of note, there were no significant differences between the S and RT+S groups in terms of baseline characteristics (all p < 0.05, Table 3). K-M survival analysis showed that there was no significant difference in the median CSS between the two groups (HR = 0.73, 95% CI = 0.30-1.77, p = 0.490, Figure 3A). A similar finding was observed in CSM (HR = 0.74, 95% CI = 0.31-1.78, p = 0.500, Figure 4A).

Effect of Preoperative Radiotherapy on CSS and CSM in Elevated CEA Subgroup
In the matched cohort, 242 patients were present with pretreatment elevated CEA levels, including 118 patients in the S group and 124 in the RT+S group. Likewise, no significant differences were observed between the S and RT+S groups in terms of baseline characteristics (all p < 0.05, Table 4). The pooled HR for the median CSS was in favor of the RT+S group as compared with the S group (HR = 0.41, 95% CI = 0.18-0.92, p = 0.032, Figure 3B), with elevated survival rates at 1, 3, and 4 years (97.12% vs. 93.22%, 90.24% vs. 79.79%, 84.91% vs. 63.51%, respectively). Similar differences were also observed between the two groups (HR = 0.42, 95% CI = 0.19-0.94, p = 0.032; Figure 4B). Furthermore, multivariate analysis in the elevated CEA subgroup showed that preoperative radiotherapy was an independent protective factor for CSS and CSM (CSS: HR = 0.36, 95% CI = 0.15-0.83, p = 0.017; CSM: HR = 0.41, 95% CI = 0.18-0.94, p = 0.036, respectively; Table 5).

DISCUSSION
The question whether preoperative radiotherapy can bring longterm survival benefit to LARC patients has been troubling the minds of surgeons and radiotherapists for a long time (7)(8)(9)(10)(11). In the current study, we found that pretreatment CEA level was a robust risk factor for prognosis after adjusting for confounding factors in different models. Furthermore, we also found that only a subgroup of LARC patients with elevated pretreatment CEA levels will benefit from preoperative radiotherapy in terms of CSS and CSM. Preoperative radiotherapy followed by radical surgery has been preferred prevalently mainly due to its advantage on downstaging, pathological complete response (pCR), sphincter preservation and superior to adjuvant radiotherapy in prevention of local recurrence (19)(20)(21), although it can (1) increase the risk of surgical complications (22,23), (2) bring radiation-related toxicity (24,25), and (3) cannot improve the long-term prognosis of LARC patients when compared with adjuvant radiotherapy (20,26,27). In this study, 87.3% of patients received preoperative radiotherapy from 2011 to 2015 in the SEER database with a satisfactory 4-year CSS rate of 88.36%, indicating the importance of standardization treatment. Furthermore, in the recent decade, more neoadjuvant strategies have been explored with inspiring results in trials of FOWARC, RAPDIO, PRODIGEL 23, and IWWD, which attach more importance to preoperative radiotherapy (7,9,28,29). The pCR rate is reported to range from 16.1% to 30% (9,(30)(31)(32),   and patients with pCR generally have a better prognosis (33)(34)(35). However, the long-term survival benefit of preoperative radiotherapy has rarely been identified in previous reports ( Table 6) (5-7, 9, 21, 28, 30, 36-38). In this study, we found that preoperative radiotherapy did not improve CSS in LARC patients before and after PSM (both p > 0.05). Additionally, we applied a competitive risk model to identify the true effect size of preoperative radiotherapy on long-term prognosis, since the rates of competition events were as high as 25% before PSM and as high as 21% after PSM. However, the results of the competitive risk model were highly consistent with the results of traditional K-M analysis, which indicated that noncancer-related mortality may have little effect on the conclusion of the study. Nonetheless, preoperative radiotherapy could not benefit LARC patients in terms of CSM before and after PSM (both p > 0.05). The reasons for this occurrence may be as follows: (1) in the era of neoadjuvant treatment followed by total mesorectal excision (TME), distant metastasis but not local recurrence is the decisive factor for longterm prognosis (7,39); otherwise, (2) in the matched cohort, apparent survival differences are observed between the RT+S and S groups both in terms of CSS and CSM but with a margin p-value (p = 0.054, p = 0.059, respectively), which indicates that a larger sample size may be needed to avoid false-negative results.
As a tumor-associated antigen, CEA has been used as a specific marker for the early diagnosis of colorectal cancer, however, its specificity is far from satisfactory (40,41). In the current study, only 1,164 (42.6%) patients presented with elevated pre-treatment CEA levels. Nevertheless, the pretreatment CEA level was found to be an independent risk factor for both CSS and CSM, and it was reconfirmed in different models by adjusting for age, sex, tumor characteristics, and treatment factors (all p < 0.05), which indicated that pretreatment CEA level was a robust prognostic factor of long-term survival, and there may be an interaction between pre-treatment CEA level and preoperative treatment on long-term prognosis.
The relationship between CEA level and preoperative radiotherapy has been explored, however, pretreatment CEA level is not an indicator of preoperative radiotherapy. Pretreatment CEA level has been identified as a predictive biomarker of neoadjuvant treatment response, as well as posttreatment CEA levels and dynamic changes in CEA levels (18). Furthermore, CEA is associated with radiation sensitivity; tumors with normal CEA are sensitive to radiation, while tumors with elevated CEA levels are resistant to radiation (12). In the current study, subgroup analysis stratified by pretreatment CEA level showed that preoperative radiotherapy would only benefit patients with elevated CEA levels. The reasons for this may be as follows: (1) preoperative radiotherapy could reduce the risk of local recurrence, which is an important risk factor for long-term prognosis; (2) the compliance of patients with elevated CEA is higher than those with normal CEA, who are much more likely to receive a full course of chemotherapy; and (3) more intensive postoperative monitoring would be conducted in patients with elevated CEA levels, indicating a more timely intervention for early recurrence/metastasis. This finding suggested that pretreatment CEA could also be used as a potential biomarker to screen patients who would enjoy the long-term survival benefit of preoperative radiotherapy.
However, there are several limitations to the current study. First, selection bias is difficult to avoid in a retrospective analysis, although a well-designed PSM was conducted in our study. Second, data on preoperative radiotherapy, including clinical target volume and radiation regimen, are unavailable in the SEER database, which would weaken the conclusion of the current study. Third, data on chemotherapy, such as regimen and courses, are also unavailable, which is one of the most important risk factors for long-term prognosis. Hence, in the present study, we excluded all those patients who had not received adjuvant chemotherapy to decrease the effect of adjuvant chemotherapy on long-term prognosis. Finally, the receipt rate of preoperative radiotherapy varies from region to region, which indicates that our conclusion needed further validation using either data outside of the USA or multicenter randomized clinical trials accordingly.

CONCLUSION
Based on our results, we conclude that pretreatment CEA level may be considered a potential biomarker to screen LACR patients who would benefit from preoperative radiotherapy in terms of long-term prognosis.

DATA AVAILABILITY STATEMENT
The dataset analyzed in this study from SEER can be obtained from: https://seer.cancer.gov/data/.

ETHICS STATEMENT
Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

AUTHOR CONTRIBUTIONS
LW, XZ, HL, LS, GC, and JW contributed to conception and design. XZ and HL conducted data collection and analyzed the data. LW, XZ, and HL interpreted the data. LW, XZ, and HL drafted the manuscript. LS, GC, and JW contributed to the critical revision of the manuscript. All authors contributed to the article and approved the submitted version.