Possible better long-term survival in left versus right-sided colon cancer – a systematic review

Colon cancer (CC) is one of the most frequent types of cancer in Denmark and the Western World. In 1990, Bufill was the first to propose that CC found in the distal and proximal location of the colon may follow different biological pathways [1]. It has subsequently been suggested that there are differences in epidemiology, perioperative course, pathology and prognosis between patients with cancers in the right and the left side of the colon [2]. The reason for this is uncertain, but it could partly be due to the different embryologic development of the two segments of the colon which may result in different molecular biological patterns of the tumours, which therefore represent two separate disease entities [3, 4]. Any such differences might have consequences for the planning of screening as well as for the treatment of patients with colorectal cancer (CRC). The aim of the present study was to perform a systematic review of the literature with a view to elucidating the subject.

METHOD

The PubMed and the Ovid databases were searched during October 2011 for literature on right- versus left-sided colon cancer. The following MeSH terms "Colonic neoplasms" and "Prognosis" were combined with "Left OR Left-sided OR left sided" and "Right OR Right sided OR Right sided". Right-sided colon cancer (RCC) is defined as malignant neoplasms in the caecum, ascending colon or transverse colon. Left-sided colon cancers (LCC) are located from the splenic flexure to the sigmoid colon, both included. The search identified 290 publications. The following inclusion criteria were then applied: 1) papers addressing the subject: right-sided versus left-sided colon cancer, 2) research papers published after 1998, 3) research papers published in English or Danish and 4) peer-reviewed publication. Additional searches were performed from the reference lists of the selected literature. After this procedure, a total of 17 studies were selected for the review (see a PRISMA flow diagram in Figure 1). Seven of these studies were prospective and ten were retrospective (Table 1) [2, 4, 6-20]. Finally, the Cochrane Database of Systematic Reviews was searched, but no relevant publications were identified. The PRISMA guidelines were followed [5].

RESULTS

The results from this systematic review are presented under the following subheadings: epidemiology, clinical manifestation, pathology and prognosis.

Epidemiology

The distribution of age and gender in the respective studies is shown in Table 2. Nine of the studies gave information about age, and it was found that the median age of patients with RCC was 71-74 years, while it was 66-71 years for LCC. Ten studies found RCC patients to be females more often than LCC patients.

Clinical manifestation

Limited data were available to determine the distribution of co-morbidity and presentation as acute versus elective operations among RCC and LCC patients. Only two articles addressed co-morbidity. One recorded co-morbidities in 85.3% of RCC patients and 82.7% of LCC patients (p < 0.01) [9]. The other study used a co-morbidity-related risk score and found that co-morbidity was significantly higher in RCC patients in stage I and III, but lower in RCC with stage II colon cancer. The risk score was found to be an independent variable for overall five-year mortality [20]. Two articles addressed presentation as an emergency and found 14% emergency presentations in RCC and 16% in LCC (p = 0.003) [10] and 12% in RCC and 15% in LCC (p = 0.014) [11], respectively.

Pathology

Union for International Cancer Control (UICC) stage, tumour-node-metastasis (TNM) status, tumour size and differentiation, and molecular biological pattern.

RCC patients were found to be in more advanced UICC stages (Stage III/IV) than LCC patients at diagnosis; and in accordance with this, RCC showed locally advanced tumour growth significantly more often (pT3/4) than LCC [2, 9, 10, 13-16, 18-20]. A larger number of lymph nodes were also harvested in RCC patients than in LCC patients, and more of the lymph nodes were cancer-positive in RCC than in LCC [2, 13, 14, 19, 20]. However, a single study painted a more complex picture with the highest proportion of lymph node-positive disease found for tumours in the caecum and at the splenic flexure [9]. One study showed that tumors in RCC were larger in size than in LCC [13], and six studies showed that tumours were more poorly differentiated in RCC than in LCC [2, 9, 13, 14, 19, 20]. Four studies found RCC to be mucinous more often than LCC [2, 6, 9, 18]. A recent study found that carcinoma of the caecum and splenic flexure had the highest proportion of lymphatic invasion, while tumours of descending colon had the lowest [9]. Further, this study showed a difference in metastatic spread between the colonic subsites, irrespective of right or left side of colon. Synchronous hepatic metastases were less frequently diagnosed in ascending and descending colon carcinoma. Metastatic spread to the lungs was most often seen in carcinoma of the caecum and sigmoid colon, whereas caecal tumours had the highest incidence of peritoneal carcinomatosis [9]. The molecular biological tumour patterns differed between RCC and LCC with more frequent C-KI-RAS mutations [3], defective DNA mismatch repair genes [4], expression of a nonfunctional p53 protein and a p53 gene mutation [6] and microsatellite instability [21] in RCC than in LCC.

Prognosis

Patients with RCC were found to have a worse overall prognosis than patients with LCC (Table 3). However, when prognosis was viewed in relation to stage, the results were a little more complex. RCC patients showed a worse prognosis in stages I and III, but not in stage II in two of the studies [2, 15]; and in three other studies [6, 13, 14], the prognosis in stage II was even better than in LCC patients. One study showed an equal prognosis in the different stages [19].

Some of the studies had further subdivided RCC and LCC, and one of these studies found that patients with cancers of the sigmoid had a better cancer-specific survival than patients with cancer in the rest of the colon [19], while patients with transverse colonic cancer seemed to have the worst prognosis of all [13, 16].

Four of the studies had performed multivariate analysis to eliminate the influence of differences in patient and tumour characteristics on the prognosis. With this approach, Suttie et al [10] showed that age, stage and mode of presentation, but not tumour location, had a significant impact on survival, while Faivre-Finn et al [11] found that stage, emergency presentation, age, as well as tumor location were significant predictors of prognosis. A third study found that location was no longer a significant factor for survival after adjusting for covariates [20]. However, in the largest prospective study included in the present review, both location, age, gender, tumour differentiation and co-morbidity were significantly related to decreased survival, but the effect of right-sided location on prognosis was found only to have an odds ratio of 1.12 (95% confidence limits: 1.018-1.226, p = 0.02) [2]. Consequently, the impact of tumour location itself on survival remains uncertain, though the different patterns in molecular biology in RCC and LCC discussed below indicate that an impact exists [4].

DISCUSSION

In an epidemiological study by Saltzstein and colleagues, increasing age was associated with a shift of anatomic site of origin of CRC from the left to right side of the colon [22]. This is in accordance with the results of the present review, where the median age at diagnosis of RCC was 71-74 years versus 66-71 years in LCC (Table 2). A report from The Danish National Board of Health stated that 40% of the patients were 75 years or older at diagnosis of CRC in Denmark (2007) [23]. In the same report, the age interval for screening was recommended to be 50-74 years. The argumentation for this was the small risk of CRC together with a reduced compliance to screening with increasing age. With this strategy, nearly half of all RCC patients and 40% of all CRC patients will not be offered screening and the accompanying advantage of early diagnosis of potential cancer, but this shall, naturally, be viewed in relation to a life expectancy in Denmark in 2008/2009 of 80.75 years for women and 76.52 years for men.

In most of the studies, a larger percentage of RCC patients than LCC patients were women (Table 2). The correlation between RCC, increased age and female gender may partly be explained by hormonal and genetic factors. The use of postmenopausal hormones may halve the risk of CRC among women [24, 25], and lifestyle and dietary habits may differ between women and men as investigated in relation to the development of serrated polyps in the right and left colon [26].

It has previously been shown that emergency resection is associated with an increased morbidity and postoperative mortality compared with elective resection [27]. This was confirmed by the findings of Suttie et al [10] and Faivre-Finn et al [11], though no difference in postoperative mortality between patients with RCC and LCC was observed in the last study. This was probably so because the differences in acute and elective operations between the two groups were relatively small.

Benedix [2] and Weiss [20] found RCC patients to have more co-morbidity than LCC patients. The significance of co-morbidity was described in a report from the Danish Cancer Research Forum from 2011 [28], where colon cancer patients diagnosed between 2007 and 2009 had a 1-year survival rate of 76% when no co-morbidity was registered, while the 1-year survival rate was only 44% in patients with a high degree of co-morbidity. These findings were confirmed by Iversen et al [29].

According to Hemminki et al [16], patients with RCC had more advanced stages at diagnosis than patients with LCC, and Snaebjornsson et al [18] found that the more advanced stages of RCC were due to tumour (T) and lymph node (N) stages, but not to metastases (M stage). RCC was associated with a larger number of harvested lymph nodes and a larger amount of positive lymph nodes among these than LCC [2, 13, 14, 19, 20]. The number of resected and positive lymph nodes is a quality parameter, as lymph node metastases are of imperative significance to prognosis and treatment [30, 31]. With regard to metastases, Benedix and colleagues [2] found that LCC more frequently spread to the liver and pulmonary systems than RCC, whereas RCC more often spread to the peritoneum, probably due to the increased prevalence of mucinous adenocarcinoma among RCC cases [2, 6, 9, 18]. There was no difference in the frequency of metastases to brain, bone, skin and/or ovaries [2]. Furthermore, several studies have shown that RCC was more poorly differentiated than LCC [2, 9, 13, 14, 19, 20] and had increased tumor size [13].

In a more recent study, Benedix and colleagues [9] found a need for a further subdivision of RCC and LCC. The study indicated that age and tumour differentiation support the common segregation into RCC and LCC, but with regard to gender, UICC stage, metastases, T- and N-status and lymphatic invasion, a subdivision into the caecum, ascending, transverse, descending and sigmoid colon is necessary. Cancers of the caecum and splenic flexure seemed more advanced (stage III/IV in the UICC classification) and more often had lymphatic invasion than cancers of the ascending and descending colon. Still, the overall picture from the present systematic literature review is that RCCs are more advanced than LCCs. Beside the hormonal and genetic factors earlier mentioned, the reason for this could be the weaker symptoms in patients with RCC than in patients with LCC. RCC is often associated with unnoticed bleeding, whereas LCC is associated with changes in bowel habits, passage trouble and obstruction [15, 20]. This may cause RCC patients to seek medical assistance later than LCC patients. Another factor that may delay the diagnosis in RCC is connected to colonoscopy and the documented inferior rate of success in the detection of RCC [32]. This is due to incomplete examinations in 3-13% of the patients and is thought to be responsible for half of all missed cancers [33]. The problem is most severe in older patients and especially in women [34], which may possibly explain part of the increased prevalence of poorly differentiated cancers in advanced stages of RCC among older females.

Overall, most of the studies found a poorer survival in RCC than in LCC. Multivariate analyses indicated that other factors than tumour location contribute to the higher mortality in RCC [2, 10, 11, 20]. These factors include age, gender, acute/elective surgery and co-morbidity, which were shown to influence the prognosis, while Benedix et al only found little impact of location itself [2]. This was supported by Weiss et al [20] who found no difference in prognosis between RCC and LCC after adjustment for age, gender, co-morbidity and postoperative adjuvant chemotherapy treatment, and by Suttie et al [10] who found age, operative intent, mode of presentation and stage to be the only variables with a significant impact on survival in multivariate analysis. However, molecular biological investigation have shown differences between RCC and LCC with more mutations of the C-KI-RAS proto-oncogene in RCC, which, in turn, was associated with a significantly poorer prognosis, thereby indicating an impact of location itself [3]. A more recent molecular biological study found that defective DNA mismatch repair (dMMR) genes were also predominantly seen in parts of the colon located orally to the splenic flexure, which, as earlier mentioned, is a part of the embryologically derived midgut (more precisely orally from the transition between the oral two-thirds and anal one-third of the transverse colon), whereas dMMR genes were rare in the hindgut-derived descending, sigmoid colon and the rectum [4]. However, patients with dMMR had a reduced recurrence rate [4]. Microsatellite instability (MSI), has also been observed more often among RCC patients than among LCC patients [21], and is similarly related to a better overall survival [35, 36] despite the fact that the effect of adjuvant chemotherapy, especially 5-fluouracil, is reduced in patients with MSI [37]. Further, Weiss et al found that a lower percentage of patients with RCC than patients with LCC completed a course of adjuvant chemotherapy, probably because of their more advanced age, which, in turn, could contribute to the lower survival rate in RCC [20].

In summary, several factors may have an impact on survival in RCC and LCC and this complex issue demands further research. The molecular biology behind RCC and LCC and potential differences in the effect of adjuvant and palliative chemotherapy and biologically targeted therapies [38] will be studied further, and the relevance of an upper limit of 74 years in the Danish screening programme needs reconsideration, as does the future potential to detect RCCs at earlier and more favourable stages. A recently initiated study on ten-year data from the nationwide Danish Colorectal Cancer database (DCCG) with a focus on differences in epidemiology, pathology and survival between RCC and LCC will hopefully further elucidate the subject.

CONCLUSION

The present review confirmed that there are clinical, pathological and prognostic differences between RCC and LCC. Patients with RCC were shown to be older, more often females, and they had more co-morbidities, more advanced tumour stages, a larger amount of harvested lymph nodes, increased tumour size, more poorly differentiated tumors, and different molecular biological tumour patterns than patients with LCC. In line with this, RCC patients were found to have a worse prognosis than LCC patients, but the reason for this seems rather complex and further research in the area is therefore needed.

Correspondence: Per Jess , Kirurgisk Afdeling, Roskilde Sygehus, 4000 Roskilde, Denmark. E-mail: pjss@regionsjaelland.dk

Accepted: 27 March 2012

Conflicts of interest:none