Colon Cancer Sidedness, Presentation, and Survival at Different Stages

Background Several prognostic factors have been used to guide therapy for colon cancer (CC). However, the relationship between CC laterality (sidedness) and prognosis remains under investigation. Objectives To assess the effect of laterality on CC presentation and survival, using a Surveillance, Epidemiology, and End Results (SEER) population-based cohort. Methods A retrospective cohort study using data from the SEER program (2007-2015). Results Of the 163,980 patients with CC, 85,779 (52.3%) presented with right-sided CC (RCC) and 78,201 (47.7%) with left-sided CC (LCC). Stage distributions were as follows: stage I, 24.1%; stage II, 27.3%; stage III, 28.2%; and stage IV, 20.4%. In an adjusted modified Poisson regression approach for risk ratio (RR), patients with LCCs were more likely to be male (RR = 1.14; 95% CI 1.12-1.15, p<0.001). As compared to stage I, stage II cancers (RR = 0.88, 95% CI 0.87-0.90, p<0.001) were less likely to be LCC. Stage IV CC was slightly less likely to be left-sided (RR = 0.98, 95% CI 0.98, 0.96-1.00, p = 0.028). The median overall survival (OS) for RCC was 87 months. The median OS for LCC was not established, as more than half of the patients diagnosed with LCC were still living at the time of the analysis. In adjusted Cox proportional Hazard model, individuals with stage I, III, and IV LCCs had superior OS as compared to those with matched-stage RCC (adjusted HR = 0.87; 95% CI 0.85-0.88, p<0.001). However, OS was worse among those with stage II disease who presented with LCC (adjusted Hazard ratio [aHR] = 1.06; 95% CI 1.02-1.11, p = 0.004). CC-specific survival (CSS) was superior for LCC versus RCC for stages III and IV but worse for II. Conclusions In this population-cohort study, LCC is associated with superior OS and CSS survival. The overall survival advantage was attributed to stage I, III, and IV disease. Individuals presenting with stage II disease exhibit superior survival if the CC is right-sided.


Introduction
Colon cancer (CC) is one of the most common malignancies in the United States and represents the second leading cause of death in the Western world [1]. A number of prognostic factors are used to guide therapy, but the value of CC laterality (sidedness) in prognosis remains controversial. The differences between the right and left colon have been hypothesized to be due to histologic, genetic, and immunologic features, all of which may confer prognostic value. Notably, the right and left colon are anatomically and embryologically different: the proximal colon is derived from the midgut and is perfused primarily by branches of the superior mesenteric artery, whereas the distal colon and rectum are derived from the hindgut and receive blood via the inferior mesenteric artery.
Several studies have explored the prognostic value of laterality with inconsistent results. Indeed, while some investigators have reported superior survival among individuals with right-sided colon cancer (RCC), others have found no difference in survival between left-and right-sided diseases [2][3][4]. One 2016 study demonstrated that RCC is associated with prolonged survival using propensity score matching [4]. However, a meta-analysis [5] of 15 studies performed that same year showed a significant survival benefit for left-sided colon cancer (LCC). Further subgroup analyses  demonstrated significant prognostic differences in Western countries. The 2016 American Society of Oncology annual meeting and the 2016 European Society of Medical Oncology annual meeting described poor survival for patients with metastatic RCC [6], especially those with RAS wild-type tumors [7,8].
These conflicting findings and previously published studies [2,3,9] have renewed our interest in investigating the effect of laterality on CC survival.

. . Study Design and Study
Population. This is a retrospective cohort study using the SEER database for identification of CC from all the registries captured in the SEER 18 program (San Francisco, Connecticut, Detroit, California, Kentucky, Louisiana, New Jersey, Greater Georgia, Hawaii, Iowa, New Mexico, Seattle, Utah, Alaska, San Jose-Monterey, Los Angeles, Rural Georgia, and Metropolitan Atlanta) who had a  . . Data Source. The SEER database is comprised of data collected by the National Cancer Institute. The SEER program collects and publishes cancer incidence and survival data using population-based cancer registries that include approximately 28% of the population of the United States. The program routinely collects data on patient demographics, tumor sites, tumor morphology, staging, surgical treatment, and follow-up.

Main Outcome Measures
Our primary outcome of interest was overall survival (OS) and colon cancer-specific survival (CSS) between right and left-sided colon cancers. Secondary outcome was the likelihood of presentation as left or right-sided cancers, for stages I-IV. The right-sided cancers were calculated using cecum, ascending colon, hepatic flexure, and transverse colon, while left-sided cancers were calculated using splenic flexure, descending colon, sigmoid colon, and rectosigmoid junction. We estimated survival in months from the date of diagnosis to the date of death for nonsurvivors; the end of the follow-up period was used to ascertain survival for survivors. Patients were stratified into three groups based on age: young (<50 years [18-49 years]), middle-aged (50-69 years), and elderly (70 years or older [70-89]).
. . Statistical Analysis. The baseline characteristics and group differences were compared using Pearson's Chi square (X 2 ) test for proportions. Nonparametric variables were compared using the Mann-Whitney-U test. The Kaplan-Meier method was used for survival analysis, and logrank test for equality of survival functions. Continuous variables were analyzed with the student t-test. Stepwise multivariable Cox regression models were built using the forward method, adjusting for baseline demographics, treatment, and tumor characteristics. Variables included in the adjusted models had a p-value <0.05 for the outcome of interest in the univariate analysis. These variables remained in the final model if they were still significant at P<0.05 in the final adjusted model, as a p-value <0.05 was deemed statistically significant in this study.
Modified Poisson approach with generalized linear model (glm) was used to estimate the risk ratio (RR) and confidence  intervals calculated by using robust error variances method [11]. Model selection was done by using Akaike information criterion (AIC) [12]. Model with smallest AIC (244872) was selected. All statistical analyses were performed using Stata version 14.2 (StataCorp, College Station, Texas, USA).   Table 2).

Results
. . Colon Cancer Laterality and Survival. The median overall survival (OS) for right-sided colon cancer (RCC) was 87 months. The median OS for that of left-sided colon cancers (LCC) could not be determined, as greater than 50% of patients with LCC were still living at the time of the analysis (Figure 2). The median cancer-specific survival was not established for LCC or RCC, as more than half of the patients included in the dataset were still living at the time of the analysis (Supplementary Figure S1). Median OS for stages III and IV was 101 and 17 months, respectively (Figure 3), while median CSS for stages IV diseases was 18 months (Supplementary Figure S2). The OS for colon cancer stages stratified by sidedness is shown in Figures  4(a) For the entire cohort, the 3-and 5-year overall survival was 70.0% and 60.2%, respectively (p<0.05). The 3-year overall survival for RCC and LCC was 67.6% and 72.5%, respectively (p<0.001), while 5-year overall survival was 58.1% for RCC and 62.4% for LCC (P = 0.003).

Discussion
Our results demonstrate that laterality has an effect on OS and CSS for both early-and late-stage CC. LCC is associated with superior OS and CSS as compared to RCC among individuals presenting with stages I, III, or IV diseases. However, for reasons that remain to be elucidated, patients presenting with stage II disease exhibited inferior OS and CSS when the primary neoplasm was located on the left side. We also noted that individuals with LCCs were more likely to be young, whereas RCCs were more common among older cohort.
Our results are consistent with those reported in a 2017 study by Lim et al. [13]. Investigators performed a retrospective analysis of 414 South Korean patients and found that patients with RCC more frequently presented with larger neoplasms and more advanced nodal disease as compared to those with LCC. Individuals with RCC also exhibited inferior 5-year OS as compared to those with LCC (82.1% and 88.7%, respectively). Our analysis revealed similar findings: the 5year OS of patients with RCC was significantly lower than that of patients with LCC at 58.1% and 62.4%, respectively.
A systematic review and meta-analysis by Petrelli et al. confirmed that LCC, as compared to RCC, is associated with a significantly reduced risk of death [14]. The Petrelli group analyzed over 1.4 million patients across 66 studies and concluded that "bearing a tumor on the left side of the colon [is] significantly associated with an absolute 19% reduced risk of death." Notably, laterality was found to have a prognostic value that was independent of stage, race, and adjuvant chemotherapy. The Petrelli group also showed that the survival discrepancy between LCC and RCC is most significant among individuals with stage IV disease. Our analysis demonstrated that the overall survival advantage of LCC was primarily due to patients with stage I, III, and IV diseases. Indeed, left-sided tumors paradoxically represented a negative prognostic factor among patients with stage II disease ( Table 4). The Petrelli group observed that the presence of microsatellite instability (MSI) was associated with a favorable outcome in stage II CC. Interestingly, for reasons that have yet to be established, stage II RCC is more likely than stage II LCC to be MSI-positive. Therefore, the prolonged survival associated with stage II RCC may be related to MSI.
It is important to acknowledge studies reporting contradictory results. In a recent population-based retrospective cohort study by Karim et al., authors used data from the province of Ontario, Canada, and found no significant difference in survival when comparing LCC and RCC and concluded that "disease laterality is not associated with longterm OS or CSS" [9]. Interestingly, however, investigators did observe that RCC was more likely to be staged as T4 and have poorly differentiated histologic features as compared to LCC; it is unclear why survival was similar between the two groups despite the more aggressive features associated with RCC. Limitations of the Karim group study include nonadjustments for confounders that represent prognostic factors in CC, such as race and ethnicity. Indeed, CC mortality rates vary significantly between different ethnic groups [15,16], and thus not adjusting for these confounders was a significant limitation of the Karim et al. 's study.
The most significant decrease in survival associated with laterality is observed in patients with stage IV disease. Our results are consistent with other studies that demonstrate markedly decreased survival among individuals with RCC as compared to those with LCC [14,17]. Indeed, this was conclusively demonstrated in two separate studies by Loupakis and Paski et al. [17,18] . The Loupakis group evaluated the association between tumor location and survival parameters in patients with previously untreated stage IV CCs receiving first-line chemotherapy ± bevacizumab in three independent cohorts: a prospective pharmacogenetics study (PROVETTA) and two randomized phase III trials, AVF2107g and NO16966. In PROVETTA, patients with LCC exhibited superior OS. This was also the case in the AVF2107g and NO16966 trials. The authors concluded that primary tumor location is an important prognostic factor in previously untreated stage IV CC.
There are several hypotheses that may explain our findings. There are significant immunological differences between the proximal and distal colon [18]. Inflammation, epithelial injury, and increased cellular permeability are most common in the proximal region of the colon [19,20]. These processes have been postulated to be due to interleukin-6 secreted by the unique microbiome present in that region of the bowel [21,22]. It is therefore conceivable that the poor prognosis observed in RCC is due, in part, to a chronic inflammatory process with consequent carcinogenesis. Indeed, some authors have hypothesized that the downstream production of proinflammatory cytokines promotes aggressive CC through increased epithelial proliferation, impaired apoptosis, and/or angiogenesis [23,24].
Microsatellite instability (MSI) colon cancers have a significantly better prognosis [25]. Right-sided colon cancers are known to have high MSI. The presence of MSI alone might not be able to explain the difference in mortality between right-sided and left-sided colon cancers. Phipps et al. [26] found increased MSI positivity in RCC, but the overall outcome and survival were still poor. Further work by Yamauchi et al. noted that frequencies of Cytosine-phosphate-Guanine (CpG) island methylator phenotype (CIMP-high), MSI-high, and BRAF mutations gradually increased from the rectum (<2.3%) to ascending colon (36-40%), followed by falls in the cecum (12-22%) [27]. The presence of BRAF mutations and CIMP-high mutations are associated with poorer prognosis [26,27]. This may explain poorer overall survival for RCC in our cohort.
Interestingly, our analysis revealed that young individuals were more frequently affected by LCC whereas RCCs were more common among the elderly. The underlying cause of the relationship between age and tumor location has yet to be established. However, increasing age represents a negative prognostic factor in colon cancer [28]. Therefore, it is possible that the poorer overall survival observed in individuals with RCC in our cohort may be related to patient age, with accompanying multiple comorbidities.
Furthermore, the poorer OS and CSS associated with RCCs may be related to screening. Indeed, several studies have demonstrated that the lower incidence and mortality in LCCs are due to relatively early diagnosis using colonoscopy [29,30]. While LCCs are more likely to present with obvious symptoms such as rectal bleeding and alteration in bowel habits leading to seeking early care, RCCs present more frequently with subtle symptoms such as microcytic anemia and weight loss which are not easily detectable until advanced stage [31,32]. Our study shows that RCCs were more likely to be T4 and advanced nodal (N) stages, and this may be related to late diagnosis.
RCCs, as compared to LCCs, were significantly more likely to be mucinous (10.7% versus 5.0%) or signet cell ring carcinoma (1.4% versus 0.7%). This is consistent with previous reports in the literature [33,34]. Mucinous adenocarcinoma produces mucin which dissects through tumor walls and promotes tumor extension; this portends a poor prognosis as well as a poor response to neoadjuvant and adjuvant chemotherapies [35,36]. Indeed, the FIRE 3 [37] and CALGB/SWOG 80405 [8] trial subgroup analysis has shown that antiepidermal growth factor receptor therapy has a decreased benefit in patients with RCC. Signet ring carcinomas are aggressive and have a propensity for extensive intramural spread as well as peritoneal carcinomatosis [38]. Consequently, these tumors are associated with an overall poor prognosis [39]. The superior OS and CSS for LCCs may therefore be due to the lower propensity for mucinous and signet ring carcinomas to develop on the left side.
Our study has some limitations. First, because of its retrospective nature we could not assess causation. In addition, the study design is inherently prone to selection bias. Second, the SEER database does not include known prognostic factors such as smoking status, diet, and obesity nor does it include baseline data on comorbidities; this may therefore be subject to residual confounding, despite multivariable analysis. The SEER database also does not have information on nonsurgical cancer directed therapies. Furthermore, tumor markers such as MSI status and BRAF, which have prognostic value, could not be determined. Despite these limitations, the major strength of this study is the large sample size, which allows for a broad and generalizable perspective on presentation and survival for CC laterality.

Conclusions
In this population-cohort study, LCC have superior OS and CSS survival. The overall survival advantage was also noted for LCC in stages I, III, and IV; however worse survival was noted for stage II. LCC is independently less likely to present as stage II and IV diseases. The findings of this study may support laterality as a prognostic indicator in considering treatment for colon cancer.

Data Availability
The data is available at https://seer.cancer.gov/data and can be accessed upon request.