Increasing evidence suggests genetic, biological and demographical difference between right and left colon cancer. Studies have also indicated age differences in the pathology of colon cancer. There is a scarcity of large-scale studies that closely examine the pathological differences regarding age and tumor location. The aim of our study was to do an extensive comparison of right- and left- sided colon cancers as well as comparing patients <50 years with older patients. A retrospective, population-based study was carried out on all patients with colon cancer in Iceland between 1955 and 2004. A total of 2293 cases were analyzed (1148 men, 1145 women). All histopathology material was re-evaluated. Differences in tumor characteristics between right and left location and younger (<50) and older (≥50) patients was evaluated in particular. Higher TNM-stage, larger tumors, vessel invasion, mucinous type, high grade and expanding tumor border occurred more frequently in right- versus left-sided lesions while annular and polypoid tumors were more common in left-sided tumors (p < 0.05). Young patients had more frequent lymph node metastases, vessel invasion, nonpolypoid lesions and infiltrating tumor border (p < 0.05). Right-sided lesions show more aggressive features, reflected in morphology and stage. Younger patients present more frequently with adverse features than do older patients. Frequency of right- and left-sided colon cancer differs by age with pronounced age-location differences in females. This supports the assumption of differences in etiology and carcinogenesis of right- and left-sided colon cancer, and between young and old patients.
Colon cancer is one of the most common malignancies in Western countries. In Iceland, colon cancer is the third most common type of cancer in men and the fourth most common in women. The age-standardized (world) incidence rate in Iceland 1998–2002 was 23.4 per 100,000 for men, and 15.6 per 100,000 for women.1
Cancer of the colon is often referred to as a homogenous entity. However, clinicians and pathologists are well aware of the diverse behavior and different morphologic features in this disease. Several publications have dealt with the issues of age and anatomic tumor subsite to identify more homogenous subgroups of colon cancer patients and compare such groups. Age differences in colon cancer have been reported with conflicting results.2, 3 Studies on the relation between right- and left-sided tumors and different variables have demonstrated that right- and left-sided lesions exhibit different genetic, biological and demographical characteristics and risk factors, suggesting that the carcinogenetic mechanism and progression of colon cancer lesions may differ with tumor location.4, 5 There is a lack of large scale population-based studies that closely examine the relationship of many different morphological tumor characteristics and stage-related factors to colon subsite and age.
Iceland has a nationwide population-based cancer registry since 1955. There are few pathology laboratories in the country and all of them have stored histological material for decades. This offered the possibility to conduct a population-based retrospective comparative study reviewing the histopathology of colon cancer diagnosed in Iceland over a 50-year period. The aim of our study was to identify clinicopathological differences in colon cancer on the basis of location and patient age.
Material and Methods
The study was approved by the Icelandic Bioethics Committee and registered by the Icelandic Data Protection Committee.
According to the Icelandic Cancer Registry, a total of 2521 colon cancer cases (colonic carcinomas) were diagnosed in Iceland between January 1, 1955 and December 31, 2004. Metachronous (not first time), nonepithelial and intramucosal colon cancers as well as cases where the origin of the primary tumor was regarded as noncolon (appendix, rectum, etc.) were excluded. Depending on available clinical and pathological information tumors on the recto-sigmoid junction were regarded as rectal according to International Union Against Cancer definition.6 In cases with multiple (synchronous) cancers, either the most advanced or the largest one in diameter was analyzed. The remaining 2293 colon cancer patients formed the basis of our study.
The median age at diagnosis was 73 years (IQR 63–80). There were 1148 men and 1145 women. Histological confirmation was present for 93% of the cases (colon resection 79%, polypectomy 1%, biopsy from primary tumor 3%, biopsy from metastasis 3% and autopsy 6%) but 7% of the cases were diagnosed without histopathological confirmation.
All hematoxylin-eosin stained slides were reviewed from the primary tumor or metastasis. The parameters noted included age at diagnosis, gender, tumor subsite within the colon, tumor size (cm), macroscopic appearance,7 annular growth, histological type,8 grade of differentiation,8 tumor border configuration,9, 10 peritumoral lymphocytic infiltration,9, 10 local extent of tumor (pT),11, 12 vessel invasion,13 lymph node status (pN),11, 12 distant metastases11, 12 and TNM-stage.11, 12 Immunohistochemistry (CK7 and CK20; DAKO-Glostrup, Denmark) was performed on all cases where histological material was only available from tumor metastasis and in cases where tumor primary in the colon was doubted. Grade was only evaluated for the histological type adenocarcinoma NOS.13 Tumors were designated right sided if they were proximal to the splenic flexure and left sided if they occurred at or distal to the splenic flexure.4 Tumor location was known for 2,133 patients (93%).
The definition of young colon cancer patients (early onset cancer) is unclear and differs by studies with a cut of point between 30 and 55 years, often depending on the specific purposes of the investigators.2, 14 In our study, age was categorized into 2 groups, young (<50) and old (≥50) as in the study of Fairley et al.3
Descriptive statistics were calculated for all variables. Categorical variables were analyzed by using the chi-square test. Continuous variables were compared using independent sample t-test and Mann–Whitney U test. The ratio of the number of cases in the right colon to the number of cases in the left colon is defined as the “R/L ratio.” Linear regression analysis was performed to investigate a possible age- and gender-specific trend in the R/L ratio. It is to be noted that an R/L ratio of greater than 1 indicates that the cancer is more common on the right than on the left. Level of significance p < 0.05 was used. Analyses were performed using the SPSS 14.0 statistical package. Differences exist in the denominator for different variables examined due to incomplete data.
The results for demographics and tumor characteristics by location are listed in Table 1. There was nonsignificant difference in the number of right- and left-sided colon cancer. Right-sided colon cancers were significantly more frequent in the older age groups (p < 0.01), but the difference in median age and gender distribution was not significant.
Table 1. Subject and tumor characteristics of colon cancer 1955–2004, overall and by location in right and left colon
Most of both the right- and left-sided tumors were of TNM Stage II (39.0% and 37.7%, respectively). In addition, there was a lower proportion of Stage I (5.7% vs. 11.7%) and higher proportion of Stage III (29.8% vs. 27.7%) and IV (25.5% vs. 22.8%) right-sided tumors compared to left-sided tumors (p < 0.01 for χ2 test of all tumor stages). Patients with left-sided lesions were diagnosed at earlier T and N stages than patients with right-sided colon cancers (p < 0.01). The median number of lymph nodes examined for right-sided cancers was 8 versus 6 (p < 0.01) for left-sided cancers. The mean number of examined lymph nodes increased from 4 (range 0–17) in the period 1955–1959 to 12 (range 0–51) in the period 2000–2004.
Histological types other than adenocarcinoma NOS including mucinous adenocarcinomas were more frequently seen in right-sided colon cancer than left-sided (13.6% vs. 6.1%; p < 0.01). Right-sided tumors were of higher grade than left-sided tumors with 22.3% of right-sided tumors being Grade III versus 4.5% of left-sided tumors (p < 0.01).
Expanding tumor border was more prevalent in the right colon than in the left colon (55.3% vs. 45.9%). Right-sided tumors were larger than left-sided cancer (median size, 5.0 cm vs. 4.0 cm). Polypoid tumor shape and annular tumor growth was more prevalent in the left colon (26.3% vs. 21.3% and 37.9% vs. 31.8%, respectively). There was no significant difference between right- and left-sided colon regarding peritumoral lymphocytic infiltration and frequency of distant metastases.
Characteristics of patients according to age are presented in Table 2. Patients under the age of 50 years with colon cancer constituted 6.2% (n = 143) of the 2293 patients. The youngest male patient was 13 years old and the youngest female patient was 18 years old. Patients in the group 13–29, 30–39 and 40–49 represented 0.4% (n = 10), 1% (n = 24) and 5% (n = 109) of all cases, respectively.
Table 2. Subject and tumor characteristics of colon cancer 1955–2004 by age groups
Patients under 50 years had significantly higher resection rate than patients 50 years and older (88.1% vs. 78.8%). Nonpolypoid-type lesions and infiltrating-type tumor border configuration were significantly more frequent in younger individuals than in the older population (84.3% vs. 75.7% and 58.9% vs. 48.7%, respectively). The median number of examined lymph nodes was higher among the younger subjects than in older individuals (9 vs. 7; p < 0.01) and younger patients had more frequently positive lymph nodes (N1 and N2) than older individuals (60.3% vs. 44.0%; p < 0.01). Vessel invasion was also more often detected in younger than older subjects (43.3% vs. 34.0%; p = 0.03).
Younger patients tended to have left-sided colon cancer, annular tumors, mucinous tumors, few well-differentiated (Grade I) tumors, less peritumoral lymphocytic infiltration, more distant metastases and more advanced stage at diagnosis although the relative difference was small (3–10%) and statistically insignificant. There was hardly any difference between cancers in the age groups regarding gender and tumor size.
Using the R/L ratio as a measure of the left-to-right shift with age, the change is shown in Figure 1. It is readily seen that with increasing age of women, increasingly more cases of colon cancer occur in the right colon. The ratio for women is around 1 at the age of 60 years. The slope of the regression line for women differs significantly from zero (slope 0.34, p < 0.01, 95% CI 0.19–0.48). The age-related shift of colon cancer is different for men with more cases occurring in the right than left colon in patients younger than 50 years. The ratio is almost 1 for men aged 50–59 and left-sided tumors are more common in men aged 60–69 after which there is a steady increase in right-sided tumors although the increase is less pronounced than in women. The slope of the regression line for men aged 60 years and older differs significantly from zero (slope 0.17, p = 0.03, 95% CI 0.05–0.30).
Through population-based examination of 2293 colon cancer patients in Iceland, reviewing retrospectively all pathology material, we have shown that there is a significant clinicopathological difference between right- and left-sided colon cancers. We have also demonstrated a difference between younger and older patients. First, right-sided colon cancer differed from left-sided colon cancer with right-sided lesions more often expressing an aggressive pathology profile, depicted both in stage and morphological variables. Second, younger patients presented more frequently with adverse pathology parameters than older patients. Third, the frequency of right- and left-sided colon cancer differs by age and gender with relatively high frequency of right-sided tumors in older women and younger men.
Increasing evidence suggests that the various differences in right- and left-sided colon cancer may partly be due to differences between the normal right and the left colon, including different embryological origins, physiological and biological properties and gene expression profiles, that predispose the tumors originating at these sites to develop along different pathways.4, 5, 15 Furthermore, accumulating evidence suggests that environmental and genetic risk factors are different for right and left colorectal tumors. For example cholecystectomy, high dietary fat intake and Lynch syndrome being related to right-sided lesions and high protein intake and familial adenomatous polyposis being related to left-sided tumors.4
Consistent with our results previous reports have indicated more frequent occurrence of mucinous phenotype16, 17 and other more rare tumor types,18, 19 high-grade tumors20–22 and nonpolypoid-type tumors23 in right-sided lesions.
Right-sided colon cancers presented at more advanced stage than left-sided tumors. This is in agreement with some previous studies17, 20, 21, 24 but contradicts with the observation of others.25, 26 Our study also showed that right-sided tumors were more advanced with regard to tumor size and vessel invasion. Larger lesions in the right colon have also been demonstrated in previous studies.20, 21, 25, 27 However, a comparison of vascular invasion between tumors in the right and left colon has rarely been performed. In this regard, we found only 1 study and it did not indicate a difference.21
When comparing right- and left-sided lesions the differences in percentages for factors with statistically significant outcome were between 5% and 17%. We are unaware of previous studies evaluating right and left differences with regard to annular tumor growth, tumor border configuration or peritumoral lymphocytic infiltration.
Some of the nonstage-related factors, which were more dominant in right-sided tumors, have been linked to worse prognosis, including nonpolypoid macroscopic appearance,28–30 high grade tumors and vascular invasion.12 Reports have both supported31, 32 and refuted19, 33, 34 the poorer prognosis of mucinous histology. Some rarer histology types, including signet ring cell carcinoma and undifferentiated carcinoma, show worse prognosis than adenocarcinoma.19, 35 Annular tumor growth, which was slightly more frequent in left-sided lesions in our study, has been related to worse prognosis.30
Our findings support the theory that tumor location is an important variable affecting tumor characteristics in colon cancer, both stage-related factors and tumor phenotype variables. The reason for right-sided colon cancer to present at higher stage is unclear but the fact that these tumors more frequently expressed more aggressive phenotype depicted in grade, type, macroscopic appearance and tumor border configuration could have contributed to the higher metastatic potential. Delay in detection might also influence the difference. Worse prognosis of right colon cancer is a matter of controversy. Some recent large studies on prognosis have indicated that right-sided lesions confer worse prognosis.20, 35, 36 Other studies have not supported this.37, 38 It is our intention to address this question in a survival study.
In our study, we also examined and compared tumor features of younger (<50) with older (≥50) patients. Colon cancer is rarely detected before the age of 50 years. In our study, patients under the age of 50 years with colon cancer constituted 6.2% (n = 143) of the 2293 patients, which is similar to what others have reported.3, 27 Variable study designs on age differences in colon and sometimes colorectal cancer make comparisons difficult and age differences have been reported with conflicting results.2 In addition, since fewer patients do undergo surgery in the older age group, as reported in some other studies36 concurrent with our findings, selection bias cannot be excluded.
Some studies have reported more frequent occurrence of advanced tumors and mucinous and high-grade phenotype in young patients. The difference in percentages is usually less than 10% although larger in some studies.2, 3, 27, 28, 39 These studies have frequently analyzed patients under the age of 40 years, which might explain why we did not confirm these results in significance testing although our current study did reveal trends toward higher percentage of mucinous adenocarcinoma and Stages III and IV tumors in patients under the age of 50 years. Although the reasons for these discrepancies remain unclear, it is possible that both environmental and genetic factors play some role. This might partly be explained by the fact that we did not grade mucinous adenocarcinomas or other rare histological types, which frequently are automatically defined as high grade.8
In our study, the younger age group had somewhat worse pathology profile than the older patient population with significantly more frequent occurrence of tumor positive lymph nodes, vessel invasion, nonpolypoid lesions and infiltrating tumor border configuration. The differences in percentages were between 8% and 16%. One study from China did not indicate any statistically significant relationship between the type of macroscopic appearance and 3 age groups (<40, 40–69 and >70).28 We did not find previous studies examining the correlation between age and vessel invasion, peritumoral lymphocytic infiltration, tumor border configuration or annular tumor growth.
Our findings of higher number of examined lymph nodes in younger patients and right-sided lesions have been reported in previous studies.20, 22, 40 This correlates well with the more frequent occurrence of vessel invasion in right colon cancer and in the patient group younger than 50 years. It is not clear whether the higher lymph node yield in younger patients and right-sided tumors, imparts higher detection rate of lymph node metastases, or if lymph nodes are more easily detected in cases with nodal metastases thus improving the lymph node harvest. Furthermore, compassion bias with regard to the younger age group cannot be excluded. The lymph node yield was relatively low considering the modern standards of minimum 12 lymph nodes studied for accurate staging, but we noted a steady increase of lymph nodes with time. There was a tendency toward more frequent occurrence of distant metastases in right-sided colon cancer and the younger patient group although this difference did not proof statistically significant.
It is of note that right-sided lesions have more frequently been related to high level of microsatellite instability (MSI)5 and the presence of MSI, both in sporadic and hereditary tumors, has been associated with improved survival.12 This might also apply for young patients since a higher proportion of those have Lynch syndrome.14
Studies using a age-period-cohort modeling of incidence rates of colorectal cancer in Connecticut (US) and Denmark indicate etiologic distinctions between cancer in the cecum and ascending colon versus in other parts of the colon in women.41 In our study, we demonstrated a gradual increase in the ratio of right to left colon cancer with age of female patients. This is consistent with several population-based reports.4, 5 In male subjects, on the other hand, right-sided lesions predominated in younger and older age groups with greater number of left colon cancer in middle age patients. The reason for this gender difference in tumor location of younger colon cancer patients is not clear.
It has been estimated that hereditary colorectal cancer syndromes account for ∼5% of cases diagnosed each year.42 Tumors of Lynch syndrome are likely to be observed in relatively younger patients and in the right colon. Familial adenomatous polyposis occurs more frequently the left colon.5, 14 In this regard, it would be interesting to know how large this group precisely is in the Icelandic population. In a population-based study from Italy,14 inherited colorectal tumors (Lynch syndrome and FAP) accounted for 38% of patients <40 years, 17% of patients 41–50 years, 10% of patients aged 51–55 and 3–4% of individuals older than 55. In an Icelandic population-based study from 2006 on familial risk of colorectal cancer in Iceland,43 the investigators observed an increased risk of colon cancer among siblings of colon cancer patients, whereas no such increase was observed for parents or offspring. Higher association in siblings has been explained by a recessive gene action or shared environment. Association in parents and offspring has been linked to dominant inherited trait like Lynch syndrome and FAP. The authors concluded that the low risk of colon cancer in parents and offspring in that study could be due to fewer dominantly inherited mutations in the Icelandic population.
The main strength of our study is that it is the only complete nationwide population based study on this subject available in the literature with a large number of patients and thorough review of all histology material, thereby minimizing the risk of any selection bias. Some potential limitations of our study require further discussion. First, selection bias can arise from cases with missing values in such a retrospective analysis. Second, especially in the earlier years of our study the assessment of distant metastases may have been less accurate than in the latter years of the research period. Third, we do not have information on the proportion of heritable colon cancer cases in our population. This would have been interesting as it would give insight into the possible impact of heritable colon cancer on the observed subsite and age differences.
In conclusion, a comparison of right and left colon cancer indicates several variations in clinicopathological features, especially that right-sided lesions more frequently display aggressive features depicted in higher stage and grade, histological type, tumor size, vessel invasion and macroscopic appearance. This possibly supports the theory of a different etiology and a possible difference in carcinogenesis of tumors at the 2 sites and confirms the importance of subsite division, for example in clinical trials. Moreover, frequency of right- and left-sided colon cancer differs by age and gender with relatively high frequency of right-sided tumors in older women and younger men. Finally, there were some disparities between younger and older patient population with more frequent occurrence of adverse factors in patients younger than 50 years of age.
We thank Professor J. Bjornsson, MD, and statistician Helgi Sigvaldason.