The Proximal Shift in the Distribution of Colon Cancer Is Independent of Age and Gender

 

 

Frank Friedenberg, MD^a

Angel Fernandez, MD^a

Barbara Sorondo, MD^b

Javid Fazili, MD^a

Leonard Braitman, PhD^c

 

Department of Medicine^a

Department of Emergency Medicine^b

Office of Research and Technology^c

Albert Einstein Medical Center

5401 Old York Road

Philadelphia, PA 19141

 

 

Key Words: cancer, distribution, age, gender, polyps

ABSTRACT

Objective: Over the past 6 decades numerous longitudinal studies have identified a proximal migration of colon cancer. This trend has been related to an actual increase in the frequency of right-sided tumors and a decrease in the frequency of left-sided lesions. The purpose of our study was to determine the recent changes in distribution of colon cancers at our institution. This information would allow a calculation of the expected sensitivity of flexible sigmoidoscopy for the detection of colon cancer.

Methods: We reviewed the results of 11,303 colonoscopic examinations consecutively performed at our institution from January 1985 to December 1998. The results of all biopsies, including those of small polyps, were reviewed and included if definitively diagnosed as adenocarcinoma.

Results: A total of 625 cases of colon cancer were identified. We found a declining percentage of colon cancers within the reach of the flexible sigmoidoscope as the years progressed from 1985 to 1998 (r = -0.10, P = .01). Changes in the age or gender distributions of study patients did not account for the observed trend.

The probability of detected colon cancers being found in the cecum increased over time (r = 0.13, P < .001). However, the proportion of colon cancers detected in the rectum was not associated with calendar time (r = 0.04, p = 0.3).

Conclusions: The current percentage of right-sided colonic neoplasms is high. Over a period of 14 years, we were able to document a continued trend of proximal migration

that was unrelated to age or gender distribution. This has important implications for policy concerning methods of colon cancer screening.

 

INTRODUCTION

Over the past 6 decades, numerous longitudinal studies have identified trends in the subsite migration of colon cancer. Prior to 1940 up to 88% of large bowel neoplasms were located in the rectum and rectosigmoid colon.¹ That figure has changed drastically, and adenocarcinoma of the rectum is now relatively uncommon.^2-4 Likewise, the incidence of carcinoma of the cecum was approximately 10% in the 1930s and 1940s but was nearly 21% in 1984.² This trend has been related to an actual increase in the frequency of right-sided tumors and a decrease in the frequency of left-sided lesions.⁵ Improved diagnosis and treatment of distal lesions and aging of the general population are unlikely to account completely for this observed proximal shift.⁵ Numerous factors have been attributed to this migration, including changes in dietary constituents,^6,7 increased rates of cholecystectomy,^8,9 changes in the prevalence of certain hereditary cancer syndromes,^5,10 and other unidentified environmental factors.  Subsite migration has important implications in the development of clinical guidelines for colon cancer screening, and the debate over sigmoidoscopy vs. colonoscopy centers around this issue.  The purpose of our study was to determine the subsite distribution of colon cancers at our institution from 1985 to 1998. In addition, we attempted to determine if the distribution of patient age or gender contributed to subsite migration as has been suggested in previous studies.^11,12

 

MATERIALS AND METHODS

We reviewed the results of all colonoscopic examinations consecutively performed at Albert Einstein Medical Center from January 1985 to December 1998. In 1985, measures were taken to ensure a complete and accurate database of procedures at our institution. Colonoscopic examinations were performed by several board-certified gastroenterologists, usually with the assistance of a senior fellow. Biopsy specimens were hand delivered to the department of pathology in labeled containers of 10% neutral-buffered formalin. If an obstructing lesion was identified, we determined whether a more proximal synchronous cancer was present from the operative report. The results of all biopsies from all colonoscopic examinations were reviewed and included if definitively diagnosed as adenocarcinoma. Retrospective data collection was easily accomplished as our clerks attach most pathology reports to the accompanying procedure report. For missing pathology reports, we utilized our hospital's computerized medical records system. If a lesion was initially read as severe dysplasia but not carcinoma and the patient subsequently had surgical resection of the lesion, the final pathology from the operation determined inclusion or exclusion. From the patients' medical records we recorded age, gender, year of the study, and location of the lesion in the colon. Racial background could not be determined retrospectively. For the purposes of this study, the colon was divided into two anatomical regions based on the reach of the 60-cm flexible sigmoidoscope. Areas within the reach of the flexible sigmoidoscope included the rectum, recto-sigmoid colon, sigmoid colon, and descending colon. Areas beyond the reach of the flexible sigmoidoscope included the splenic flexure, transverse colon, hepatic flexure, ascending colon, and cecum.

 

STATISTICAL ANALYSIS

Spearman rank correlation coefficients (r) were determined between the year the cancer was detected and the percentage of lesions found (1) within the reach of the flexible sigmoidoscope, (2) in the cecum, and (3) in the rectum. A logistic regression model was used to predict the probability that a detected cancer was within the reach of the sigmoidoscope given the year the cancer was detected (from 1985 to 1998). To investigate if age or gender was responsible for the observed association, they were controlled for by adding them to that model. Logistic regression models were tested for goodness of fit using the Hosmer-Lemeshow test. Statistical analyses were performed using SPSS V. 9.0 (SPSS Inc., Chicago, IL) and Stata version 6.0 (College Station, TX).

 

RESULTS

We reviewed a total of 11,303 consecutive colonoscopic examinations. After elimination of incomplete examinations (due to poor bowel preparation or other technical reasons) and examinations revealing no more than benign pathology, a total of 625 colon cancer cases were identified. This represented 5.5% of all examinations. If a colonoscopy was incomplete, we were not able to determine if an alternative study was performed (i.e., contrast barium enema). No patient had malignancy in more than one location of the large bowel. There were 329 women (52.6%) and 296 men (47.3%). The age range was 30 to 97 years, with a mean (+SD) of 71.7 (+11.6).

We hypothesized that the probability of finding colon cancers in the cecum would increase over time and the corresponding probability of finding cancers in the rectum would decrease over time. The percentage in the cecum did increase over time from 1985 to 1998 (r = 0.13, P < .001). However, the proportion of colon cancers detected in the rectum was not associated with calendar time (r = 0.04, P = .3; Figure 1). Controlling for the age and gender of patients left those findings unchanged.

An important clinical question is whether the probability of detected colon cancers within reach of the flexible sigmoidoscope decreased from 1985 to 1998. In our study group, there was an downward trend in the percentage of colon cancers within reach of the flexible sigmoidoscope as the years progressed from 1985 to 1998 (r = -0.10,

P = .01; Figure 2).

Using a logistic regression model to smooth the data, the probability of the detected cancer being reached by a flexible sigmoidoscope decreased with the year of cancer detection from 65% in 1985 to 50% in 1998 (P = .02). The downward trend over time was unchanged when age and gender were controlled for by including them in the model. Thus the decreasing percentage of detected cancers within reach of a sigmoidoscope from 1985 to 1998 cannot be accounted for by the age or gender distribution of such cancer patients at our hospital during this time period.

 

DISCUSSION

Our data provide current information concerning the anatomic distribution of colon cancer. They suggest that the prevalence of proximal colonic neoplasms remains high. Prior to data analysis, we divided the colon into those segments within and beyond the reach of a 60-cm sigmoidoscope. We found a migration of cancers from 1985 to 1998, with <48% within reach of the sigmoidoscope between 1996 and 1998 at our institution.  In our analysis, we assumed all hypothetical sigmoidoscopic examinations would have reached the descending colon. In practice, sigmoidoscopic examinations are often limited due to poor patient tolerance, inadequate preparation, and anatomic problems such as severe diverticulosis. If one assumes that only 50% of sigmoidoscopic examinations are completed to 60 cm,¹³ then the sensitivity of this test for detecting colon cancers may be substantially lower than predicted by our data.

Our study has several important limitations. All patients referred for colonoscopic examination presumably had symptoms or a significant family history. Our study group does not represent the normal screening population. In a healthy population, the subsite distribution of colon cancer may be different from our symptomatic patient group. A second limitation of the study is reliance on the endoscopist to identify the subsite location of the neoplasm. Conceivably, a lesion designated as "splenic flexure" may actually have come from the descending colon or vice versa. It is unlikely that skilled endoscopists would have frequently mislabeled the subsite location. Because such possible misclassification bias is not systematic, it would likely not bias our results substantially. A final limitation is that we did not record whether a distal polyp was present in those cases in which a malignancy was found proximal to the reach of a sigmoidoscope. Presumably, these patients would have had a subsequent colonoscopy to search for a proximal lesion.

The removal of distal colon polyps during prior sigmoidoscopy in our population would have influenced the subsequent distribution of colon cancer.  However, it is common practice to remove polyps only at the time of colonoscopy at most institutions, including our own. Although we do not know how many of our cancer patients underwent prior flexible sigmoidoscopy (likely to be a very low number), none of the patients had a sigmoidoscopy with polypectomy prior to diagnosis of cancer.

Our data adds to the growing body of literature concerning trends in subsite location of colon cancer in the United States. A Medline search identified numerous articles documenting the same trend in the proximal migration of colon cancer starting around 1930.^1-3,12,14-17 Prior to our study, Obrand and Gordon presented the most recent data, which included the period of 1979 to 1994.⁴ During that time period, the percentage of rectal cancers dropped from 22% to 12%. The corresponding percentages of lesions beyond the transverse colon increased from 22% to 31%, paralleling the trends seen previously.⁴

Several studies throughout the world have attempted to identify similar trends in subsite cancer distribution. Data from Canada and New Zealand showed trends which appear to parallel those occurring in the United States.^18,19 However, age-standardized incidence rates of colorectal cancer at different subsites in Switzerland and in ethnic Chinese citizens residing in Singapore have remained unchanged, at least over the past 25 years.^20,21

Racial background may also play an important role in observed trends in the incidence and subsite distribution of colon cancers.²² Demers and colleagues demonstrated a substantial proximal migration of colon cancers in African Americans in a Detroit population over the period of 1973 to 1994. This trend was most notable in African American women.²³ A study involving an indigent population in Florida demonstrated that the percentage of proximal neoplasms in African Americans was 53% vs 40% for whites over a 27-year period beginning in the 1960s.²⁴ Substantially different results were published by Devesa and Chow after analysis of data from the Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute.²⁵ Their data, which encompassed the years 1976 to 1987, demonstrated no racial difference in the rates of proximal neoplasms in African Americans in comparison to white patients.²⁵ Again, information concerning race was not available in our study.

Because polyps are considered to be the precursors of colonic adenocarcinomas, a change in their distribution would be expected to parallel the changes in colon cancer distribution. In a study from 1950 to 1989, Offerhaus and coworkers were able to show a proximal migration of polyps at Johns Hopkins.²⁶ Surgical data encompassing the time period of 1960 to 1980 demonstrated similar results.¹⁷ Likewise, Bernstein and colleagues looked at the distribution of colon polyps identified by barium radiograph and found a dramatic shift in the percentage of proximal polyps in those older than 60 years.²⁷ The review by Levi and coworkers of the Vaud Cancer Registry raises questions concerning this hypothesis. Although there was a substantial proximal migration of benign polyps when comparing the time periods of 1979 to 1983 with 1995 to 1996, there was no identified shift in the subsite distribution of adenocarcinomas.²¹ 

Characteristics of the population, such as age and gender, may play an important role in the distribution of colon cancer. However, neither the age or gender distribution of

patients in our study accounted for subsite migration. No other studies have provided sufficient data to draw definitive conclusions regarding the influence of age and gender on proximal migration trends. However, several studies have addressed the association of age and gender with the location of colon cancer without considering trends over time. Slater and colleagues, utilizing data from the Connecticut Tumor Registry, demonstrated that the mean age of individuals with right-sided adenocarcinoma was approximately 3 years greater than those diagnosed with left-sided lesions.¹² Nelson and coworkers, using the Illinois Tumor Registry, demonstrated that unadjusted incidence rates for proximal colorectal cancer increased with age regardless of gender, up to age 85.¹¹ Data from Slattery and colleagues, using 3 large databases, demonstrated that patients older than 70 years were more likely to have proximal neoplasms, although this association was more evident in women.²⁸ Similarly, Alley and Lee found that more patients older than 70 years presented with proximal neoplasms predominantly due to the later age of presentation of women with right-sided lesions.⁸ Other studies have confirmed that the formation of proximal neoplasms is more likely in women, especially with increasing age.^29-31

An explanation for the shift in subsite distribution of colon cancer has been the focus of intense investigation. Cholecystectomy, an increasingly common procedure in the latter half of the twentieth century, causes a subsequent rise in the flux of bile acids through the colon and the production of secondary bile acids predominantly in the cecum. From animal data, bile acids have been shown to enhance epithelial cell proliferation and stimulate DNA synthesis in premalignant colonocytes.³² The secondary bile acid, deoxycholate, appears to be particularly responsible for this phenomenon.³²

McMichael and Potter summarized the results of 20 studies looking at the relationship between cholecystectomy and colonic neoplasia. Of 20 studies, 14 demonstrated an increase in the risk of colon cancer in association with prior cholecystectomy. Half showed an increased risk of proximal cancers.³²

Dietary constituents may also influence the development of proximal colonic neoplasia. A nested case-control study, which reviewed the intake of calcium in men of Japanese ancestry living in Hawaii from 1900 to 1919, identified an inverse relationship between dietary calcium and the subsequent risk of sigmoid colon cancer. Dietary fat had no influence on subsite cancer development.⁷ Hirayama's study (from Japan) found no association between the consumption of milk, meat of any type, green or yellow vegetables, rice or wheat, or soybeans or cigarette smoking and the development of proximal bowel cancers. However, he did find a strong relationship between the consumption of alcoholic beverages and sigmoid colon cancer.³³ In summary, the role of diet in the development of proximal colonic neoplasms remains unclear.

Certain genetic alterations in the colonic epithelial, such as those occurring in hereditary cancer syndromes, predispose individuals to the development of proximal neoplasms.^5,10 Although understanding of the molecular changes that take place and the ability to screen this population improve, individuals with proximal neoplasms represent a small segment of the population of patients with colon cancer and improved diagnosis is unlikely to impact overall trends substantially. 

 

CONCLUSION

Results from our institution show that the incidence of right-sided colonic neoplasms is high. In fact, over a period of 14 years, we were able to confirm a trend of proximal migration identified in other studies. That this trend was independent of age and gender at our institution suggests that these factors do not account for the observed trend in the general population. The percentage of colon cancers reachable by sigmoidoscope is less than 50% in recent years at our institution. This may have important implications for policy concerning methods of colon cancer screening. The etiology of this migration remains unclear. Convincing data regarding the role of previous cholecystectomy and dietary constituents are lacking.

 

REFERENCES

1.      Cady B, Stone MD, Wayne J: Continuing trends in the prevalence of right-sided lesions among colorectal carcinomas. Arch Surg 128:505-509, 1993.                                             

2.      Ghahremani GG, Dowlatshahi K: Colorectal carcinomas: Diagnostic implications of their changing frequency and anatomic distribution. World J Surg 13:321-325, 1989.

3.      Rosato FE, Marks G: Changing site distribution patterns of colorectal cancer at Thomas Jefferson University Hospital. Dis Colon Rectum 24:93-95, 1981.

4.      Obrand DI, Gordon PH: Continued change in the distribution of colorectal carcinoma. Br J Surg 85:246-248, 1998.

5.      Bufill JA: Colorectal cancer: Evidence for distinct genetic categories based on proximal or distal tumor location. Ann Intern Med 13:779-788, 1990.

6.      Willett WC, Stampfer MJ, Colditz GA, et al: Relation of meat, fat, and fiber intake to the risk of colon cancer in a prospective study among women. N Engl J Med 323:1664-1672, 1990.

7.      Stemmerman GN, Nomura A, Chyou PH: The influence of dairy and non-dairy calcium on subsite large-bowel cancer risk. Dis Colon Rectum 33:190-194, 1990.

8.      Alley PG, Lee SP: The increased risk of proximal cancer after cholecystectomy. Dis Colon Rectum 26:522, 1983.

9.      Todoroki I, Friedman GD, Slattery ML, et al: Cholecystectomy and the risk of colon cancer. Am J Gastroenterol 94:41-46, 1999.

10.  Delattre O, Law DJ, Remvikos Y, et al: Multiple genetic alterations in distal and proximal colorectal cancer. Lancet 353-355, 1989.

11.  Nelson RL, Dollear T, Freels S, et al: The relation of age, race, and gender to the subsite location of colorectal carcinoma. Cancer 80:193-197, 1997.

12.  Slater G, Papatestas AE, Tartter PI, et al: Age distribution of right and left-sided colorectal cancers. Am J Gastroenterol 77:63-66, 1982.

13.  Maule WF: Screening for colorectal cancer by nurse endoscopists. N Engl J Med  330:183-187, 1994.

14.  Kee F, Wilson RH, Gilliland R, et al: Changing site distribution of colorectal cancer. Br Med J 305:158, 1992.

15.  Beart RW, Melton LJ, Maruta M, et al: Trends in right and left-sided colon cancer. Dis Colon Rectum 26:393-398, 1983.

16.  Maglinte DD, Keller KJ, Miller RE, et al: Colon and rectal carcinoma: Spatial distribution and detection. Radiology 147:669-672, 1983.

17.  Greene FL: Distribution of colorectal neoplasms: A left to right shift of polyps and cancer. Am Surg 49:62-65, 1983.

18.  Mamazza J, Gordon PH: The changing distribution of large intestinal cancer. Dis Colon Rectum 25:558-562, 1982.

19.  Jass JR: Subsite distribution and incidence of colorectal cancer in New Zealand 1974-1983. Dis Colon Rectum 34:56-59, 1991.

20.  Huang J, Seow A, Shi CY, et al: Colorectal carcinoma among ethnic Chinese in Singapore: Trends in incidence rate by anatomic subsite from 1968 to 1992. Cancer 85:2519-2525, 1999.

21.  Levi F, Randimbison L, Vecchia CL: Trends in the subsite distribution of colorectal carcinomas and polyps. Cancer 83:2040-2042, 1998.

22.  Winawer SJ, Fletcher RH, Miller L, et al: Colorectal cancer screening: Clinical guidelines and rationale. Gastroenterology 112:594-642, 1997.

23.  Demers RY, Severson RK, Schottenfeld D, et al: Incidence of colorectal adenocarcinoma by anatomic subsite. An epidemiologic study of time trends and racial differences in the Detroit, Michigan area. Cancer 79:441-447, 1997.

24.  Johnson H, Carstens R: Anatomical distribution of colonic carcinomas: Interracial differences in a community hospital population. Cancer 58:997-1000, 1986.

25.  Devesa SS, Chow W: Variation in colorectal cancer incidence in the United States by subsite of origin. Cancer 71:3819-3825, 1993.

26.  Offerhaus GJA, Giardiello FM, Tersmette AC, et al: A shift from distal to proximal neoplasia? Four decades of adenomatous polyps at the Johns Hopkins Hospital. Gastroenterology 98(supplement):A301, 1990.

27.  Bernstein MA, Feczko PJ, Halpert RD, et al: Distribution of colonic polyps: Increased incidence of proximal lesions in older patients. Radiology 155:35-37, 1985.

28.  Slattery ML, Friedeman GD, Potter JD, et al: A description of age, sex, and site distributions of colon carcinoma in three geographic areas. Cancer 78:1666-1670, 1996.

29.  Jensen OM: Different age and sex relationship for cancer of subsites of the large bowel. Br J Cancer 50:825-829, 1984.

30.  Stewart RJ, Stewart AW, Turnbull P, et al: Sex differences in subsite incidence of large bowel cancer. Dis Colon Rectum 26:658-660, 1983.

31.  Vobecky J, Leduc C, Devroede G: Sex differences in the changing anatomic distribution of colorectal carcinoma. Cancer 54:3065-3069, 1984.

32.  McMichael AJ, Potter JD: Host factors in carcinogenesis: Certain bile-acid metabolic profiles that selectively increase the risk of proximal colon cancer. JNCI 75:185-191, 1985.

33.  Hirayama T: Association between alcohol consumption and cancer of the sigmoid colon: Observation from a Japanese cohort study. Lancet 2:725-727, 1989.