Colorectal carcinoma (CRC) represents one of the most common health burdens worldwide. The histological classification of CRCs was according to the World Health Organization classification system. Non-mucinous adenocarcinoma (NMC) is the most common histological subtype, followed by mucinous adenocarcinoma (MC), and signet-ring cell carcinoma (SRCC). In contrast to NMC, MC and SRCC are adenocarcinomas in which the cancer cells produce excess mucin. MC accounts for 10–15% of CRC and is composed of extracellular mucin pools that cover more than 50% of the lesion [1,2]. SRCC, which accounts for 1% of CRC, is defined as signet-ring cells making up more than 50% of the tumor area. In addition, CRC that contains both mucinous and signet-ring cell components is currently classified as MC if extracellular mucins make up more than 50% of the tumor [1,2]. The histological subtype is considered to play a role in CRC patients with different clinical characteristics and survival outcomes. Though the prognosis of MC is still controversial [3–9], SRCC proved to be associated with impaired outcome in almost all previous studies [10–18]. Because tumor location and tumor stage have been related to the prognosis, these variables could have acted as potential confounders in previous researches. To assess a prognostic role of histological subtype independent of tumor location and stage, it is necessary to control these key factors strictly.

In this study, we analyzed the prognostic value of histological subtype in CRC, using the data from the Surveillance, Epidemiology, and End Results (SEER) database, which included a large number of patients from the United States national population-based data. We hypothesized that the prognostic value of each histological subtype might differ in different tumor locations and tumor stages of CRC.

In most of the relating studies, mucin-secreting adenocarcinoma (including MC and SRCC) and right-sided colon cancer were associated with adverse prognosis [10–12,19]. The occurrence of these histological subtypes may differ widely among different geographic areas. For SRCC, according to the literatures, the frequency could vary from 1% in the United States to 18.5% in Jordan and Lebanon [19,20]. In this study, the proportion of colorectal SRCC in the SEER database has been rising during the past decades. The frequency of SRCC in the right colon was much higher than in the left colon and the rectum in each period, which was consistent with previous reports. In these studies, right-sided colon cancer was considered to have something to do with worse prognosis compared with the left-sided colon cancer and RC [16]. The higher frequency of SRCC in right-sided colon cancer may also contribute to the poorer outcome of this group.

This study confirmed a number of recognized characteristics associated with SRCC, such as younger age at diagnosis, more patients with single tumor, more advanced tumors, and poorer tumor grade. Additionally, our research showed that colorectal SRCC patients were less likely to achieve a CDS and more often be assigned for radiotherapy than MC and NMC in RC. Consistent with the study of Vallam et al. [21], in which patients with SRCC had a higher rate of circumferential resection margin (CRM) positivity (19%) than patients with non-SRCC (4%), making a CDS impossible. In a previous study concerning RC [22], we also observed that the patients in SRCC group were more likely to receive preoperative radiotherapy (PRT) compared with patients in the MC and NMC groups, but SRCC remained a factor of poor prognosis in locally advanced stage (stages II and III) RC patients who were in a PRT setting. Both phenomena might be attributed to advanced stage at diagnosis for this special entity. Besides, a predilection to occur in the right colon, larger tumor size, and a greater number of lymph nodes examined were common features between SRCC and MC.

Although colorectal SRCC and MC were considered to be associated with poor prognosis in many previous studies, reports on the prognosis value of SRCC and MC stratified by tumor location and stage are still rare. From our results, SRCC had the worst CSS in all patients and at each stage of each tumor location, except for stage I of left-sided colon cancer. One of the reasons for the poor prognosis of SRCC may be advanced stage at diagnosis, as mentioned above. The higher CRM-positive rate in SRCC made a CDS impossible, leading to a higher risk of local recurrence and worse survival [10,21,23,24]. Low differentiation at diagnosis, associated with high risk of vascular invasion and lymph node involvement [17,24], may be another explanation for the adverse prognosis of SRCC. Furthermore, in their study, Ho-Su and colleagues [25] presented a higher incidence of scirrhous carcinoma in the SRCC group than the NMC group, which might lead to symptoms of intestinal obstruction. Bowel obstruction may prevent the endoscope to pass beyond the tumor, and finally can affect the rate of missed diagnosis of synchronous carcinoma in colorectal SRCC. Besides, in the same research, they found that most patients with SRCC showed an invasive growth pattern, which had been shown to be an independent prognostic factor in patients with stages I-III CRC. In other studies [11,23], the decreased expression of adhesion molecules (such as E-cadherin, β-catenin) and the amplification of Bcl-224 were related to reduce cell-cell adhesion, loosen the surrounding structure of tumor cells and enabled them to spread far away or form peritoneal metastases, which could not be treated by radical surgery and lead to poor prognosis.

Though MC has been reported to have some similar clinicopathological features compared with SRCC, the prognostic value of MC is still controversial [3–9]. Hugen et al. [6] reported that the poor prognosis for MC was only seen in RC. Gao et al. [4] found that MC was associated independently with poorer prognosis for RC and was an independent protective survival indicator in right-sided colon cancer. In our study, MC acted as a favorable factor in right-sided colon cancer, in keeping with Peng’s research. The mechanism could potentially be related to the interaction of MC with other site-specific factors. One of the hypotheses is that hereditary non-polyposis colorectal cancer (HNPCC) and high-frequency microsatellite instability (MSI-H) tend to be associated with right-sided of colon cancer, and usually occur in MC. HNPCC and MSI-H have proven to be associated with good response to adjuvant therapy and improved disease free survival [26–29]. Therefore, the protective effect of MC in right-sided colon cancer may be due to the higher proportion of HNPCC and MSI-H in this tumor location group. In a study of colorectal MC, Chung-Ta et al. [30] found that MCs without both serrated adenocarcinoma (SAC) morphology and CpG island methylator phenotype (CIMP) positive status exhibited 3.955 times greater risk of cancer relapse than MCs having both characteristics or either one, and SAC was associated with proximal location and CIMP-positive status, which may be another explanation for the phenomenon. Despite all these, further research is still needed to confirm this phenomenon and elucidate the mechanism behind it.

In the 5-year survival analysis, there were no differences between MC and NMC in stages I and III of right colon, stage I of left colon, and stage IV of rectum. The intrinsic mechanism to explain this phenomenon has not been reported.

In consistency with previous findings, the current study also showed that SRCC subtype, male gender, age ≥ 70 years, tumor size > 5 cm, advanced stage, poor differentiation were all significant factors in the deterioration of CRC survival. In contrast, multiple tumor numbers, accepted CDS, higher numbers of examined lymph nodes (≥ 12), administration of radiotherapy were associated with better survival. Notably, MC histological subtype proved to be an independent protective factor of prognosis in right colon, but when it appeared in left colon or rectum, an opposite effect was observed. Our findings suggested that clinical decision should take these factors into account.

Our study data included a large number of patients from the United States, using national population-based data, avoiding the biases associated with single-institution experience or limited sample sizes. However, due to the non-random nature of SEER, some limitations of the current study deserved comment. First, in the case of a large population size data, it was not feasible to review the individual pathological diagnosis. Due to the heterogeneity of pathological information and the diversity of pathologist interpretations, there may be misclassifications of SRCC and MC in the SEER database. Because tissue samples were not available, we were unable to determine the histological type of each patient in this study. In order to explore possible heterogeneity between registries, we compared the percentages of SRCC and MC between each center and found no significant differences. Secondly, information regarding chemotherapeutic treatment of patients was not available in SEER data set. Furthermore, information on indicators which might be related with patient prognosis (such as HNPCC, MSI, SAC, CIMP, etc.) has not been recorded in our study. Finally, data on recurrence time and patterns were unavailable in this research. CSS is a reasonable surrogate of CRC-specific outcome.

In conclusion, SRCC had the worst survival among the three histological subtypes of CRC in our study. MC was associated with favorable prognosis in right-sided colon cancer but not in other tumor locations.