Introduction

Colorectal cancer (CRC) is an important public health problem; it is the third commonly diagnosed cancer and the second leading cause of death worldwide [1]. CRC is a significantly heterogeneous cancer on the basis of its histological type, grade, stage, and treatment response. Different genetic, etiological, environmental, microbiotic, and lifestyle factors lead to the heterogeneity of CRC [2] and influence its prognosis. Advancements in medical therapy have gradually improved the survival of patients. These advancements include the exploitation and utilization of new drugs, improved treatment, and discovery of predictive factors. Emerging studies have demonstrated that primary tumor location can serve as an important predictive factor, which might predict curative effects in metastatic CRC [3, 4]. The outcomes of patients with left-sided cancers are better than those of right-sided cancers in metastatic CRC [5,6,7,8]. However, the available information and findings remain limited and contradictory in localized CRC [9, 10]. Thus, we aimed to evaluate the clinical characteristics and prognosis of PTL in CRC.

Materials and methods

Study design and patient selection

This population-based cohort study analyzed the correlation between PTL and outcomes in CRC. The study data were extracted from the Surveillance, Epidemiology, and End Results (SEER), which covered 27.8% patients with cancer in the USA [11]. SEER*Stat version 8.3.5 was employed. We examined data from Incidence—SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2017 Sub (2000–2015). The data were extracted and described according to data items and codes as documented by the North American Association of Central Cancer Registries (NAACCR) [12]. We extracted data for all cases of CRC between 2006 and 2015, which were coded according to THE year of diagnosis (NAACCR Item390). Primary cancer site and histology were coded using the criteria in the third edition of the International Classification of Diseases for Oncology (ICDO-3) [13]. The patients who were diagnosed at autopsy or only by death certificate and without histologically confirmed cancer (NAACCR Items 490 and 2180) and with occurrence of another malignancy preceding CRC (NAACCR Item 380) were excluded. These analyses were further stipulated to patients with adenocarcinoma identified by the ICDO-3 histology codes 8140, 8144, 8210,8211, 8220, 8221,8255, 8260, 8261, 8262, and 8263; mucinous 8480; mucin-producing adenocarcinoma 8481; and signet ring cell carcinoma 8490 (NAACCR Item 522). These analyses were also stipulated to patients with clear stage (0, I, IIA, IIB, IIIA, IIIB, IIIC, and IV) identified by the DERIVED AJCC-6 STAGE GRP (NAACCR Item 3000). Right-sided colon cancers included C18.0-cecum, C18.2-ascending colon, C18.3-hepatic flexure of colon, and C18.4-transverse colon. Left-sided CRCs included C18.5-splenic flexure of colon, C18.6-descending colon, C18.7-sigmoid colon, C19.9-rectosigmoid junction, and C20.9-rectum (NAACCR Items 522 and 523).

Statistical analysis

Statistical analyses were carried out using SAS statistical software (SAS, version 9.4; SAS Institute Inc.). Chi-square test was used to evaluate proportions. Multivariable logistic analysis was conducted to analyze the correlation between every factor with overall survival (OS).The Cox regression model and Kaplan–Meier method were employed to analyze the correlation between PTL and OS. A two-sided p value < 0.05 was considered as statistically significant.

Results

Study population

We identified a population-based sample of 311,239 patients diagnosed with CRC between 2006 and 2015. We excluded living patients without survival time information from OS analysis. Thus, 248,861 patients were retained in this cohort. The selection process for patients in the study is listed in Fig. 1.

Fig. 1
figure 1

Selection process for patients in the cohort study

Full size image

Population characteristics

The stage distribution was 3.8, 25.1, 25.8, 26.8, and 18.6% for stages 0, I, II, III, and IV, respectively. The characteristics of patients are listed in Table 1. This study comprised more patients with left-sided cancers than those with right-sided cancers [57.0% (177,444 of 311,239) vs. 43.0% (133,795 of 311,239)]. Females were more likely to present with right-sided cancers (53.3% [71,330 of 133,795] vs. 42.9% [76,182 of 177,444], P < 0.001) than left-sided cancers. Patients with right-sided tumors were older [≥ 75 years, 43.4% (58,027 of 133,795) vs. 25.9% (45,916 of 177,444), P < 0.001] than those with left-sided tumors. Compared with left-sided tumors, right-sided tumors were more likely to develop into T3 cancers [50.0% (66,877 of 133,795) vs. 44.8% (79,545 of 177,444), P < 0.001], T4 cancers [15.8% (21,084 of 133,795) vs. 12.3% (21,753 of 177,444), P < 0.001], mucinous or mucin-producing adenocarcinoma [10.8% (14,413 of 133,795) vs. 5.0% (8869 of 177,444), P < 0.001], and signet ring cell carcinoma [1.4% (1918 of 133,795) vs. 0.7% (1314 of 177,444), P < 0.001].

Table 1 Demographics and clinical characteristics of patients with colorectal cancer between 2006 and 2015
Full size table

Factors correlated with survival

The factors correlated with 5-year survival are listed in Table 2. Multivariable analysis indicated that the following factors were correlated with inferior prognosis (OS): male, old age (≥ 75 years), unmarried status, histopathology grades 3 and 4, mucinous adenocarcinoma, mucin-producing adenocarcinoma, signet ring cell carcinoma, stages III and IV, and right-sided tumor.

Table 2 Factors correlated with five-year overall survival among 248,861 patients with colorectal cancer
Full size table

Exploratory analyses of the associations between PTL and survival

We performed exploratory analyses to identify the association between PTL and the OS of the patients with CRC. The patients of all stages (stages 0–IV) were merged in analysis to identify the prognostic relevance of PTL. The PTL was correlated with prognosis. Cox regression analysis revealed that patients with right-sided tumors showed inferior OS (56.1% vs. 60.2%), and the hazard ratio was 1.224 (95% CI, 1.208–1.241, P < 0.001) in stages 0–IV. The Kaplan–Meier analysis results listed in Fig. 2 also indicated that the patients with right-sided tumors had inferior OS. We further analyzed the correlation between PTL and the OS in every stage. Stage-specific Cox regression analysis revealed that patients with right-sided cancers had inferior OS in every stage (Table 3). Patients with stage III showed the greatest difference between the left-sided and right-sided cancers in OS (64.4% vs. 54.0%, P < 0.001) than patients with other stages.

Fig. 2
figure 2

Patients with left-sided tumors had better 5-year overall survival (60.2% vs. 55.2%, P < 0.001) compared with right-sided tumors in stage 0–IV

Full size image
Table 3 Primary tumor location correlated with 5-year overall survival among 248,861 patients with colorectal cancer
Full size table

Discussion

In this study, we evaluated the clinical characteristics and prognostic relevance of PTL in CRC. We further confirmed through historical data that patients with right-sided cancers were more likely to be old, female and mucinous adenocarcinoma or signet ring cell histology than those with left-sided cancers [14,15,16]. Our data were consistent with the growing body of evidence showing that those with left-sided cancers have better prognosis than patients with right-sided cancers among patients with metastatic CRC (stage IV) [5,6,7,8]. Our analysis further showed that patients with left-sided cancers exhibited better prognosis than patients with right-sided cancers in all stages (including localized CRC). When grouped based on tumor stage, all patients with left-sided cancers showed significantly better prognosis than those with right-sided cancers regardless of stage. This difference was not apparent between patients with metastatic CRC and those with localized CRC. However, the reason for the different survival rates of patients with different PTL in CRC remains unclear. We speculate that differences in embryological origin and detection time may contribute to this discrepancy. The embryological junction between the midgut and hindgut leads to a potential watershed area in the area of the splenic flexure, and it is supplied by the superior and inferior mesenteric arteries. The rectum also arises from the hindgut and the blood supply from the inferior mesenteric artery. So, we included it in the left half colon analysis. Due to the larger diameter of the right-sided colon tube, all patients with right-sided cancers had later onset of clinical symptoms, such as abdominal pain and intestinal obstruction. Right-sided tumors were more likely to develop into advanced cancers (stage III and IV). After controlling for tumor stage and histology, the patients’ prognosis in left-sided cancers remained better than those of right-sided cancers. This result also suggested that the difference in prognosis between the two groups may be related to genetic and environmental factors. Right-sided cancers are more likely to involve genome-wide hypermethylation and hypermutations than left-sided cancers [17,18,19,20]. Gene analyses elucidated four biological consensus molecular subtypes (CMSs) in CRC [21]. Notably, the differential CMSs were distributed between the left-sided and right-sided cancers, and “microsatellite unstable/immune” CMS1 and “metabolic” CMS3 were more prevalent in right-sided cancers than in left-sided cancers [21]. The different PTLs also exhibited varied microbiota and histories of exposure to potential carcinogenic toxins [22, 23]. In stage IV, patients with left-sided cancers have a higher rate of liver metastases and lung metastases was found when compared with those of right-sided cancers, whereas patients with right-sided cancers were associated with a higher rate of peritoneal metastases and metastases at other sites [24]. In fact, several recent studies have exhibited that PTL may be prognostic and predictive of the response to antiepidermal growth factor receptor (EGFR) therapy in mCRC. Trials using cetuximab as an anti-EGFR therapy, including CRYSTAL and FIRE-3, showed that the prognosis of patients with left-sided cancers were superior to those of patients with right-sided cancers [25]. In terms of treatment, compared to right-sided cancers, some patients with rectal cancers received preoperative or postoperative radiotherapy. Some studies demonstrated that those treatments could improve local control [26,27,28,29]. However, those studies also demonstrated that OS were not improved [27,28,29,30]. It might not affect the results (OS) of our study that the rectum was incorporated into the left colon for analysis. In addition, it might affect the OS that patients with right-sided tumors were older than those with left-sided tumors. Therefore, the clinical characteristics and prognostic mechanisms of different sites of colorectal cancer require further study, especially on locational cancers. Our results also differed from other study [10]. Such different results might be due to the different definitions for right- and left-sided cancers. Cancers located from the rectum to the splenic flexure colon are commonly defined as left-sided cancers, whereas those from the splenic flexure colon to the cecum are defined as right-sided cancers [8, 17, 31, 32]. By contrast, others employed different definitions of left-sided cancers, which only included the descending and sigmoid colon cancers, and right-sided cancers, which only included the cecum and ascending colon cancers; moreover, they excluded patients with cancers in other colorectal locations [10]. Our study was based on the former definition. Such different definitions may explain the inconsistency between the results of the two studies.

Limitations and strengths

To our knowledge, our study is the largest work to analyze the influence of PTL on the prognoses of patients with CRC in all stage. However, the study had some limitations. Data on treatments, family history, performance status, and molecular features were unavailable in the SEER database. Despite these limitations, the present study had some strengths. First, the population-based nature of the registry is associated with a high degree of generalizability. Second, our study reported data over a 10-year period and included over 311,000 patients with CRC. The large sample size was also associated with a high degree of power. Third, we used different analysis measures to prove the results of this study.

Conclusions

This study revealed the clinical characteristics and prognostic value of PTL in patients with CRC. The OS of patients with left-sided cancers was better than that of patients with right-sided cancers regardless of stage. We strongly recommend regarding PTL as a stratification factor in future studies. We encourage developing clinical and translational studies to elucidate the causative relationship between PTL and prognosis, and developing strategies for the prevention measures and clinical management of CRC by stratifying on the basis of PTL.