In this large population-based cohort study, we discovered that 1) SR patients had the worst OS of the three subtypes of CRC; 2) RT significantly increased the risk of death among patients with adenocarcinoma CRC, particularly among patients who were elderly, male, and had a tumor in the colon; and 3) RT significantly reduced the risk of death in SR patients, particularly among patients who were older, male, and with the tumor located the rectum.
The current study's findings of poor prognosis in SR patients were in line with the majority of previous research. In a Korean study[5], the five-year survival rate in SR patients was only 26.8%, which was significantly worse than MC (58.1%, P < 0.001) and adenocarcinoma (62.9%, P < 0.001). By analyzing 244,794 CRC patients from the National Cancer Data Base (NCDB), Hyngstrom JR et al. found patients with signet ring cells colorectal adenocarcinomas had poorer outcomes compared with non–signet ring adenocarcinomas of either the colon or rectum[11]. The association of MC with the risk of mortality of CRC patients was controversial. According to a population-based study in Singapore[7], MC is an independent prognosis factor for poorer outcomes of CRC-related death. Whereas, other studies found that poor MC outcomes were more related to tumor stage or specific tumor sites, suggesting that MC is not an independent prognostic factor[20, 21].
RT was discovered to be a risk factor for adenocarcinoma patients' prognosis in this study. Subgroup analyses further confirmed the elevated risks of RT for the prognosis of adenocarcinoma patients were observed among patients with colon cancer, however, not with rectum cancer. Several previous studies partly supported our results. A randomized controlled trial (RCT) involving 1861 CRC patients conducted by Kapiteijn E et al. found the overall survival rate among eligible patients at two years was 82.0% in the group receiving both radiation and surgery and 81.8% in the group receiving surgery alone (P for difference = 0.840)[22]. Also, results from the randomized multicenter TME trial observed there was no improvement in survival of 10-year OS for adenocarcinoma patients (53.3% for surgery alone vs. 50.5% for surgery after radiotherapy; p for differences = 0.679)[17]. Unfortunately, no existing studies found that RT elevated the risk of death among adenocarcinoma patients with colon cancer. However, the advantages of RT should be noted as well. RT could reduce the incidence of local recurrence, down-stage locally advanced cancer, and increased resection rate for tumors, as well as raise the likelihood of anus reservation[23–25].
We did not find a significant association between RT and OS in MC patients which was similarly shown in a prior study. Importantly, RT significantly lowered the risk of death for SR patients, particularly when the tumor was situated in the rectum rather than the colon. Consistent with our findings, Ling and his colleagues found that RT was an independent prognostic factor associated with improved survival in stage III rectal SR, rectal patients with stage III underwent preoperative RT had a better prognosis than patients who received surgery alone (HR = 0.47, 95% CI: 0.27–0.79)[26]. Furthermore, Xu Guan et al. discovered that RT might improve survival in patients in stages II and III by evaluating data from 1808 rectum cancer patients in the SEER database[18], which is corroborated by another research by San-Gang Wu[27].
We provided evidence that RT is closely associated with worse prognosis among adenocarcinoma patients and better prognosis in SR patients was not yielded to analytical methods that consider the competing risk of cancer-free death. The occurrence of the event of interest is frequently prevented by another event in time-to-event studies, making it difficult to observe the occurrence of the event of interest. During an average follow-up of 27 months, 4166 people died from causes other than CRC. Compared to the results of Cox regression, the results were slightly strengthened in the competitive risk approaches, confirming the primary conclusions of the current study. In the current study we applied both the CS models and the SD models. Despite comparable conclusions, the CS model and SD model are distinct. The CS model would be more appropriate for investigating the causal association. The SD model would be more suited to predicting the likelihood of the result[19].
The underlying mechanism remains uncertain. However, several mechanisms may help to explain our findings. First, the expression of the p53 protein was found to be higher in the adenocarcinoma component of the tumor than in the signet ring carcinoma component[28]. The presence of wild-type p53 in cancer cells has been linked to susceptibility to radiation or chemotherapy-induced damage, whereas mutant p53 has been linked to radio- and chemotherapeutic resistance[29]. Second, SR cancer showed significantly negative/lower expression of epidermal growth factor receptor (EGFR), compared with other pathologic types of CRC[30]. EGFR regulates a variety of physiological responses, including cell proliferation, apoptosis, and differentiation, and is suspected to have a role in a variety of cancers. In in vitro tests of malignant cell research, overexpression of EGFR has been linked to radio-resistance[31, 32]. Third, the expression of cyclooxygenase (COX)-2 was reduced in SR cancer compared to other colorectal cancer subtypes. COX-2 overexpression was found to be more likely to demonstrate a poor response to chemoradiotherapy [33]. Overall, the high expression of p53, EGFR, and COX-2 in adenocarcinoma results in poor response to chemoradiotherapy, and may confer early and late toxicity from RT, affecting patients' long-term health-related quality of life.
The current study's strength is that it offers a unique perspective on the possible link between RT and the prognosis of various CRC histological subtypes. Other strengths included the enormous amount of data on CRC patients, which allows for accurate extrapolation of the findings. In addition, competing risk regressions were used to confirm our findings, which improved the study's accuracy. On the other hand, the limitations must be acknowledged. First, SEER, in particular, is lacking in precise data on chemotherapy and RT. It does not distinguish between "unknown" and "no." Second, there was no information on the patient's comorbidities, family history, body measures, clinical exams, or laboratory testing, which might have been associated with CRC outcomes and survival. Third, targeted therapy and immunotherapy are becoming more common in clinical practice for the treatment of CRC, However, dur to the limited data, we could not assess the effect of targeted therapy and immunotherapy on the association of RT with the risk of CRC-related death. Lastly, pathologic data was collected from a variety of hospitals and not submitted to a centralized review.