Gender Differences in ColorectalPolyps and TumorsTimothy M. McCashland, M.D., Randall Brand, M.D., Elizabeth Lyden, M.S., Pat de Garmo, R.N., A.P., andCORI Research ProjectDepartments of Medicine and Preventive and Society Medicine, University of Nebraska Medical Center,Omaha, Nebraska; and Oregon Health Sciences University, Portland, Oregon

OBJECTIVES: To use a national endoscopy database (CORI)to determine 1) whether gender differences are noted in theprevalence and location of polyps and tumors; 2) whetherwomen have a higher rate of right-sided polyps or tumors;and 3) whether age influences these results.

METHODS: CORI database from April 1, 1997 to February19, 1999, captured in a computer-generated report, wasanalyzed. Polyps for this study were defined as sessile orpedunculated and as.9 mm. Tumors were defined as le-sions characteristic of adenocarcinoma (mass, apple-core).Pure right-sided colon (PRS) was defined as cecum, ascend-ing, hepatic flexure; right-sided as PRS plus the transversecolon; and left-sided as the splenic flexure, descending,sigmoid and rectum.

RESULTS: Men have a greater risk of polyps [odds ratio(OR), 1.5] and tumors (OR, 1.4) than women. The risk offinding polyps and tumors at colonoscopy increases withage, with the highest risk noted in those.69 yr of agerelative to patients,50 yr of age (polyps, OR5 2.7;tumors, OR5 4.0). Right-side polyps and pure right-sidedpolyps as defined by the study design were noted to be morefrequent than left-sided polyps in patients.60 yr of age.Women have a greater risk of developing pure right-sidedpolyps (OR, 1.2), tumors (OR, 1.6) and right-sided tumors(OR, 1.5) than men.

CONCLUSIONS: Men have a higher prevalence of colon pol-yps and tumors than women. A progressive risk of polyp ortumor formation is noted with aging. Women had a greaternumber of pure right-sided polyps and tumor development.Colonoscopy is needed to correctly diagnose an increasingprevalence of right-sided pathology in the elderly. (Am JGastroenterol 2001;96:882–886. © 2001 by Am. Coll. ofGastroenterology)

INTRODUCTION

Gender differences in prevalence and location of colorectalpolyps and tumors have not been studied extensively. Re-ports have noted a lower rate of colorectal adenomas, butsimilar rates of colon cancer, in women as compared withmen (1, 2). Single-center experiences (1) and tumor registry

(2) were used in the past to analyze gender differences. Adiverse multicenter experience with large numbers of examshas not previously been reported. The purpose of the studywas to use a national endoscopy database, the ClinicalOutcomes Research Initiative (CORI), to determine 1)whether gender differences are noted in the prevalence andlocation of polyps and tumors; 2) whether women have ahigher rate of right-sided polyps or tumors; and 3) whetherage influences these results.

MATERIALS AND METHODS

CORI is a nationwide endoscopy report generator and da-tabase; its purpose is to measure clinical outcomes related toendoscopy. It consists of a physician network of 62 centersand 542 endoscopists equally distributed throughout thecountry. Endoscopists from private-practice groups (n537), academic medical centers (n5 18), and VeteransAdministration hospitals (n5 8) comprise the researchgroup.

Data were captured in a computer-generated endoscopyreport and transmitted to a central database for analysis.Colonoscopy reports from 4/1/97 to 2/19/99 were queried.The index colonoscopy was used for each patient; duplicateexams on the same patient were not included in the studydatabase. Polyps were defined as sessile or pedunculatedpolyps .9 mm; colonoscopy reports with polyps,9 mmwere excluded; size was by endoscopist estimation. Agree-ment between endoscopists on size of polyps or measuringdevices were not available for this database. Tumors weredefined as characteristic mass or apple-core lesions sugges-tive of adenocarcinoma. Histological and epidemiological(diet, weight, use of nonsteroidal anti-inflammatory drugs,calcium intake, family history, family history of polyps/GItumors, or personal history of polyps) data were not con-sistently retrievable with this database. However, review of87 patients with polyps and 26 patients with tumors meetingstudy criteria was done at one site (University of NebraskaMedical Center) comparing endoscopic findings with histo-logical data. Histological findings in 87 polyp patients in-cluded the following: 41 with tubular adenoma, 28 withtubulovillous adenoma, seven with tubulovillous adenoma

THE AMERICAN JOURNAL OF GASTROENTEROLOGY Vol. 96, No. 3, 2001© 2001 by Am. Coll. of Gastroenterology ISSN 0002-9270/01/$20.00Published by Elsevier Science Inc. PII S0002-9270(00)02431-X

with high-grade dysplasia, two with tubulovillous adenomawith carcinoma in situ, five with hyperplastic polyp, twowith colitis, and one each of lipoma and hamartomatouspolyp. Histological review of patients thought endoscopi-cally to have tumor (n5 26) revealed 17 with adenomacarcinoma, seven with tubulovillous adenoma with high-grade dysplasia, one with Crohn’s disease, and one withsarcoma. Pure right-sided lesions (PRS) were defined ascecum, ascending, hepatic flexure, right-sided as PRS plusthe transverse colon and left-sided as the splenic flexure,descending, sigmoid and rectum. Patients were separatedinto age groups as follows:,50, 50–59, 60–69, and.69 yrof age.

Logistic regression was used to evaluate whether age,gender, and/or an interaction of age and gender were pre-dictive of the development of polyps or tumors among allpatients who had colonoscopy in this database. It was alsoused to evaluate whether these variables were predictive ofthe development of right- or left-sided polyps among pa-tients who had polyps and right- or left-sided tumor forthose with tumors. Age was grouped by decades (50–59 yr,60–69 yr, and.69 yr) in the regression models for ease ofinterpretation; the group of younger-aged patients (,50 yr)was the reference group. There was no statistically signifi-cant interaction between age and gender in any of the

regression models, so the results are presented based onmodels including only age and gender as main effects. In allmodels, the overall test of age was highly significant, and sopairwise comparisons of the age groups were done withoutadjustment of thep values. Ap value of ,0.05 was con-sidered statistically significant.

RESULTS

Patient CharacteristicsPatient characteristics and locations of polyps and tumorsare summarized in Table 1. The number of colonoscopies inboth gender groups was large. Increasing numbers of pa-tients in each decade were noted. A higher number of menwas noted in all age categories, except the one of age,50yr, in which equal numbers of men and women were seen.The top five indications/evaluations for patients with polyps,in descending order, were as follows: history of polyps,hematochezia, positive fecal occult blood test, abnormalflexible sigmoidoscopy, and anemia. Similar indications/evaluations was noted for patients with tumors, albeit in adifferent order: hematochezia, anemia, positive fecal occultblood test, abnormal flexible sigmoidoscopy and the com-bination of anemia and hematochezia. No differences inindications/evaluations were noted between the genders.

Table 1. Patient Characteristics

Characteristics

Polyps (n5 3,268) Tumors (n5 675) Total (n5 30,727)

No. % No. % No. %

GenderMale 2149 65.8 440 65.2 17,268 56.2Female 1119 34.2 235 34.8 13,459 43.8

Age (yr),50 311 9.5 51 7.6 5975 19.450–59 621 19.0 95 14.1 6954 22.660–69 979 30.0 181 26.8 7638 24.9691 1357 41.4 348 51.6 10,160 33.1

Location of polyps and tumorsCecum 365 11.2 99 14.7Ascending 518 15.9 91 13.5Hepatic flexure 156 4.8 58 8.6Transverse 207 6.3 39 5.8Splenic flexure 60 1.8 21 3.1Descending 390 11.9 36 5.3Sigmoid 1205 36.9 173 25.6Rectum 367 11.2 158 23.4

Right- vs left-sided locations*Right 1041 31.9 277 41.0Left 2227 68.1 398 59.0

Pure right-sidedvs left-sided locations†Pure right 782 23.9 237 35.1Left 2486 76.1 438 64.9

* Right sided5 cecum, ascending, hepatic flexure and transverse colon; left sided5 splenic, descending, sigmoid, and rectum colon.† Pure right sided5 cecum, ascending, and hepatic flexure; left sided5 transverse, splenic, descending, sigmoid, and rectum.

883AJG – March, 2001 Gender Differences in Colorectal Polyps and Tumors

Relationship of Age/Gender With PolypsMen had more polyps than women (OR, 1.52; 95% CI,1.41–1.64). A progressive risk of development of polypswith each decade of life was seen with the largest risk inthose. 69 yr of age (Table 2). In patients with polyps, therewas a statistically significant difference between age groups50–59 yr and 60–69 yr (OR, 1.73vs2.58,p , 0.0001). Nodifference was noted between age groups 60–69 yr and.69 yr.

Right Versus Left-Sided PolypsComparisons of the ORs for the age categories for right- andleft-sided polyps showed an increase with aging (Table 2).With each decade of aging, the risk increased (age 50–59vsage 60–69 yr,p 5 0.004; age 60–69vs age.69 yr, p 50.0007). Similar results were found with defined pure right-sidedversusleft-sided polyps (age 50–59vsage 60–69 yr,p 5 0.0008; age 60–69vs age.69 yr, p 5 0.0008).

Women had a greater risk of developing pure right-sidedpolyps than did men (OR, 1.2; 95% CI, 1.01–1.42). How-ever, this gender difference was lost when the transversecolon was included in the definition of rightversus leftcolon.

Relationship of Age and Gender With TumorThe relationship of age and gender and location of tumors issummarized in Table 3. Men had more tumors than women

(OR, 1.43; 95% CI, 1.22–1.68). Similar to the case withpolyps, the risk of tumors increased with each decade, withthe largest risk in those.69 yr of age (Table 3). Compar-ison of each decade in relation to tumors revealed signifi-cance (age 50–59vsage 60–69 yr,p , 0.0001; age 60–69vs age.69 yr, p , 0.0001).

Right Versus Left-Sided TumorsWomen had more defined right-sided and pure right-sidedtumors than men (Table 3). Again as noted in polyps, rightand pure right-side tumors increased with aging. Evaluationof right and pure right-sided tumors again increased withaging, (right sided, age 50–59vsage 60–69 yr,p 5 0.007;pure right-sided, age 50–59vsage 60–69 yr,p 5 0.0007).However, no significant differences were seen between thegroup aged 60–69 yr and those.69 yr in regards to rightand pure-right lesions.

DISCUSSION

Colorectal cancer is the second leading cause of death(50,000/yr) in the United States (3). The use of colonoscopyto diagnosis and clear polyps has been effective at reducingboth the incidence and mortality of colorectal cancer (4).Gender differences in prevalence and location of polyps andtumors have not been studied extensively in the past. The

Table 2. Relationship of Age and Gender to the Presence of Polyps

Characteristics

Polyps Right Sided Pure Right Sided

OR 95% CI OR 95% CI OR 95% CI

GenderFemale‡ 1.0 1.0 1.0Male 1.52* (1.42–1.64) 0.94 (0.81–1.11) 0.82† (0.69–0.97)

Age (yrs),50‡ 1.0 1.0 1.050–59 1.73* (1.50–1.99) 1.04 (0.75–1.43) 0.93 (0.64–1.31)60–69 2.58* (2.26–2.95) 1.44† (1.08–1.94) 1.43† (1.03–1.98)691 2.73* (2.41–3.11) 1.95* (1.47–2.59) 1.97* (1.44–2.70)

Odds ratios (OR) and 95% confidence intervals (CI) are displayed for all predictor variables. Statistically significant predictors are indicated with † for p , 0.05 and * forp ,0.001.

‡ Reference group.

Table 3. Relationship of Age and Gender to the Presence of Tumors

Characteristics

Tumors Right Sided Pure Right Sided

OR 95% CI OR 95% CI OR 95% CI

GenderFemale‡ 1.0 1.0 1.0Male 1.43* (1.22–1.68) 0.67† (0.48–0.92) 0.62† (0.44–0.86)

Age (yr),50‡ 1.0 1.0 1.050–59 1.56† (1.50–1.99) 0.72 (0.34–1.53) 0.73 (0.33–1.61)60–69 2.73* (1.99–3.73) 1.53 (0.79–2.98) 1.47 (0.74–2.93)691 4.02* (2.99–5.40) 1.98* (1.05–3.71) 1.79* (0.93–3.44)

Odds ratios (OR) and 95% confidence intervals (CI) are displayed for all predictor variables. Statistically significant predictors are indicated with † for p , 0.05 and * forp ,0.001.

‡ Reference group.

884 McCashland et al. AJG – Vol. 96, No. 3, 2001

CORI endoscopy database provides a unique ability to studythe differences in polyps and tumors in relationship to ageand gender because it incorporates a diverse selection ofpatients from all regions of the country and practice settings.This allows a fair representation of the population as awhole to study. However, these results may only pertain topatients undergoing colonoscopy with the noted indications/evaluation of this study and may not reflect the prevalenceof these lesions for indications of screening purposes.

Different methodologies have previously been used tostudy gender differences in relation to colorectal polyps/tumors (1, 2). Nelsonet al. used the Illinois tumor registryin a population-based epidemiological study from 1986 to1991. Male incidence rates for colorectal cancer locatedanywhere in the colon, adjusted for age and race, weregreater than incidence rates for women (OR5 1.32 forproximal lesions and 1.68 for distal lesions) (2). Our find-ings were similar in that men had a higher prevalence oftumors; however, women were noted to have more rightedtumors. Noshirwaniet al. used a single-center (ClevelandClinic) endoscopy database reporting a lower prevalence ofadenomas in women, similar to our findings. Conversely,they noted that women were significantly less likely thanmen to have right-sided adenomas; the exception was in the70- to 79-yr age group (1). Our findings are similar in thatolder individuals had more right-sided polyps but at an evenlower age: $60 yr of age. A major strength of theNoshirwani et al. study was the inclusion of histologicaldata. Polyps.9 mm were chosen to increase the likelihoodof adenomatous polyps and reduce the inherent miss-rate ofpolyps. Prior reports have noted a miss-rate of 0–6% inadenoma.1 cm (5). The review of one CORI site’s histo-logical data produced results consistent with this prior re-port: 10% of polyps at the University of Nebraska were notadenomatous. Endoscopic correlation with tumors was bet-ter: 92% were adenocarcinoma or tubulovillous adenomawith high-grade dysplasia. Unfortunately, the CORI data-base is unable to document the histology of polyps or tumorsfor all contributing members. African-American womenwere more likely than whites to develop proximal colorectaltumors, as reported by Nelson’s study (2). Regrettably, racewas also unavailable for all procedures in the CORI database.

Is there a clinical significance of right-side pathologyversusleft-side disease? Elsalehet al., in a recent study ofadjuvant chemotherapy in Dukes’ C colon carcinoma, re-ported that patients with right-side tumors showed a signif-icant survival benefit with chemotherapy (48vs 27%),whereas those with left-sided tumors had only a minimalbenefit (6). In addition, women had a better response thandid men. This report also reported that right-sided tumorswere more common in women than in men (58vs 42%);however, specific age group analysis was not provided.

Why do men have a higher prevalence of polyps andtumors? One theory is that estrogen may have a protectiverole in prevention of polyp formation by mechanisms ofestrogen receptor genes (7), decreased secondary bile acid

production (8), and decreased serum levels of insulin-likegrowth factors (9). Why do older women have more right-sided neoplasms than men? When women age, reducedestrogen production may alter the bile acid composition,resulting in more toxic secondary bile acids, exposing theproximal colon to these neoplastic promotors. Foleyet al.recently showed that malignant colon tissue had a selectiveloss of estrogen receptorb protein expression when com-pared with normal colon tissue in the same patient. Thisstudy suggests that the protective effects of estrogen againstcolon cancer may be mediated by estrogen receptorb post-transcriptional regulatory mechanism involved in the ex-pression of this receptor (10). Grodsteinet al., using meta-analysis of 18 epidemiological studies of postmenopausalhormone therapy, found a 20% reduction in risk of coloncancer and a 19% decrease in the risk of rectal cancer forpostmenopausal women who had ever taken hormone ther-apy, compared with women who never used hormones (11).Interestingly, duration of treatment had no further risk re-duction. In the Nurses’ Health Study, postmenopausal hor-mone therapy was associated with a marked reduction ofproximal colon cancers (12).

The clinical implications of our results would push forcolonoscopy for older patients, because of the right-sidedpathology. A recent study by Rexet al. favors colonoscopyfor all average-risk patients because the majority of patientswith proximal cancer do not have concomitant polyps/can-cer distal to the splenic flexure (13).

In conclusion, our endoscopy database study found mento have a higher prevalence of polyps and tumors thanwomen. The risk of polyps and tumor development in-creases with age; the greatest risk is to those.60 yr of age.Women had more pure right-sided polyps and more rightand pure right-sided tumors.

ACKNOWLEDGMENTS

CORI is supported by grants of the American Society forGastrointestinal Endoscopy, Bard Interventional Products,and Astra Pharmaceuticals. The data in this manuscript wereobtained from the Clinical Outcomes Research InitiativeNational Endoscopic Database (CORI-NED), with supportfrom National Institutes of Health (NIDDK) grant UO1-DK57132–01.

Reprint requests and correspondence:Timothy M. McCash-land, M.D., University of Nebraska Medical Center, 983285 Ne-braska Medical Center, Omaha, NE 68198-3285.

Received Mar. 23, 2000; accepted Nov. 6, 2000.

REFERENCES

1. Noshirwani KC, van Stolk RU, Beck GJ. Gender variation incolorectal adenoma prevalence. Am J Gastroenterol 1998;93:A347.

2. Nelson RL, Dollear T, Freels S, et al. The relation of age, race,

885AJG – March, 2001 Gender Differences in Colorectal Polyps and Tumors

and gender to the subsite location of colorectal carcinoma.Cancer 1997;80:193–7.

3. Landis SH, Murray T, Bolden S, et al. Cancer statistics 1998.CA Cancer J Clin 1998;48:6–29.

4. Winawer SJ, Zauber AG, Ho MN, et al. Prevention of colo-rectal cancer by colonoscopic polypectomy. N Engl J Med1993;329:1977–81.

5. Rex DK, Cutler CS, Lemmel GT, et al. Colonoscopic missrates of adenomas determined by back-to-back colonoscopies.Gastroenterology 1997;112:24–8.

6. Elsaleh H, Joseph D, Grieu F, et al. Association of tumour siteand sex with survival benefit from adjuvant chemotherapy incolorectal cancer. Lancet 2000;355:1745–50.

7. Issa JPJ, Otaviano YL, Celano P, et al. Methylation of theestrogen receptor CpG island links aging and neoplasia inhuman colon. Nat Genet 1994;7:536–40.

8. Everson GT, McKinley C, Kern F. Mechanisms of gallstoneformation in women. Effects of exogenous estrogen and di-

etary cholesterol on hepatic lipid metabolism. J Clin Invest1991;87:237–46.

9. Campagnoli C, Biglia N, Altare F, et al. Differential effects oforal conjugated and transdermal estradiol on insulin-likegrowth factor I, growth hormone and sex hormone bindingglobulin serum levels. Gynecol Endocrinol 1993;7:251–8.

10. Foley E, Jazaeri A, Shupnik M, et al. Selective loss of estrogenreceptorb in malignant human colon. Cancer Res 2000;60:245–8.

11. Grodstein F, Newcomb PA, Stampfer MJ. Postmenopausalhormone therapy and the risk of colorectal cancer: A reviewand meta-analysis. Am J Med 1999;106:574–82.

12. Grodstein F, Martinez ME, Platz EA, et al. Postmenopausalhormone use and the risk of colorectal adenoma and cancer.Ann Intern Med 1998;128:705–12.

13. Rex DK, Chak A, Vasudeva R, et al. Prospective determina-tion of distal colon findings in average-risk patients withproximal colon cancer. Gastrointest Endosc 1999;49:727–30.

886 McCashland et al. AJG – Vol. 96, No. 3, 2001