GASTROENTEROLOGY 2006;130:19411949

CLINICALALIMENTARY TRACT
Predictive and Protective Factors Associated With Colorectal
Cancer in Ulcerative Colitis: A Case-Control Study
FERNANDO S. VELAYOS,* EDWARD V. LOFTUS Jr, TINE JESS, W. SCOTT HARMSEN,
JOHN BIDA, ALAN R. ZINSMEISTER, WILLIAM J. TREMAINE, and WILLIAM J. SANDBORN
*Division of Gastroenterology and Hepatology, University of California San Francisco, San Francisco, California; Division of Gastroenterology
and Hepatology, Mayo Clinic College of Medicine, Rochester, Minnesota; and Department of Medical Gastroenterology C, Herlev University
Hospital, Herlev, Denmark

Background & Aims: Predictive and protective factors

associated with colorectal cancer in chronic ulcerative
colitis are not well described. Surveillance colonoscopy
and 5-aminosalicylic acid therapy may mitigate cancer
risk, but there is debate because these variables have
not been evaluated in the same study. The presence of
postinammatory pseudopolyps and use of other antiinammatory medications may be important variables
that inuence risk, but data are sparse. Methods: Variables associated with colorectal cancer were registered
in 188 patients with ulcerative colitis-related cancer and
matched controls. Conditional logistic regression, adjusted for age at colitis diagnosis and colitis duration,
identied a nal set of variables independently associated with colorectal cancer. Results: In the nal multiple
variable model, the most important factors associated
with colorectal cancer were a history of pseudopolyps
(OR, 2.5; 95% CI: 1.4 4.6), 1 or 2 surveillance colonoscopies (OR, 0.4; 95% CI: 0.2 0.7), smoking (OR, 0.5;
95% CI: 0.2 0.9) and use of corticosteroids (OR, 0.4;
95% CI: 0.2 0.8), aspirin (OR, 0.3; 95% CI: 0.1 0.8),
nonsteroidal anti-inammatory drugs (OR, 0.1; 95% CI:
0.03 0.5), and 5-aminosalicylic acid agents (OR, 0.4;
95% CI: 0.2 0.9), although the latter was not statistically signicant after 5 years. Primary sclerosing cholangitis and immunosuppressive use were not statistically
signicant. Conclusions: These results suggest that, in a
population matched for extent and duration of chronic
ulcerative colitis, surveillance colonoscopy and use of antiinammatory medications may reduce the risk of colorectal cancer. A history of postinammatory pseudopolyps
appears to be a predictive factor for cancer.

he risk of colorectal cancer (CRC) in chronic ulcerative colitis (CUC) depends on a balance of predictive and protective factors. Evidence suggests that
chronic inflammation promotes the carcinoma sequence
in CUC.1 Even with this knowledge, it remains uncertain
why, in a group of CUC patients with similar extent and

duration of colonic inflammation, some will develop

CRC, but most will not. By investigating the importance
of several variables for CRC within the same study
population, important cancer prevention strategies and
high-risk subgroups may be identified.
The importance of several well-known predictive and
protective factors associated with CRC is not well described. This may be in part because they have not been
evaluated within the same study. Many studies are small,
making it difficult to study several variables simultaneously and to adjust for potential confounders. Early age
onset of CUC,2,3 family history of CRC,4 7 sulfa allergy,6
primary sclerosing cholangitis (PSC),8,9 clinical activity,
and backwash ileitis10 have been reported to be predictive of cancer risk in some, but not all, study populations.6,1114 Regular physician visits,6 cigarette smoking,4 folic acid,4 surveillance colonoscopy,15 and chronic
5-aminosalicylic acid (5-ASA) therapy6,14 have been reported to be protective in some, but not all, studies.16 18
The effects of 5-ASA therapy and surveillance colonoscopy on cancer risk have not been evaluated within the
same study. With respect to 5-ASA therapy and surveillance colonoscopy, prospective trials could provide level
A evidence, but are unlikely to be performed,19,20 leaving
high-quality observational studies (level B evidence) as
the main source of evidence guiding patient care.
The importance of other potentially important factors
associated with CRC is also not well described. Postinflammatory pseudopolyps, a common and benign endoscopic finding that reflects underlying inflammation and
regeneration, was recently reported to be a potentially
Abbreviations used in this paper: 5-ASA, 5-aminosalicylate; CRC,
colorectal cancer; CUC, chronic ulcerative colitis; NSAID, nonsteroidal
anti-inammatory drug; PSC, primary sclerosing cholangitis.
2006 by the American Gastroenterological Association Institute
0016-5085/06/$32.00
doi:10.1053/j.gastro.2006.03.028

1942

VELAYOS ET AL

important predictive factor for cancer; however, this

finding has not been confirmed. 5-ASA, a common antiinflammatory therapy used in CUC, has been reported to
reduce the risk of CRC14,21,22; however, few studies have
extended this finding to ask a more global question of
whether use of other anti-inflammatory medications,
both prescription and over-the-counter, may also reduce
this risk.
Few centers have a sufficient number of CUC-associated cancers to evaluate multiple variables in a single
study. We therefore performed a large retrospective casecontrol study among 188 patients with CUC and CRC
and 188 control patients matched for extent and duration
of CUC with the following aims: to identify the most
important predictive and protective variables associated
with cancer risk; to quantify the effect of surveillance
colonoscopy and 5-ASA therapy on the development of
cancer; and, finally, to identify independent risk factors
for CRC in CUC that remain after adjusting for surveillance colonoscopy and anti-inflammatory therapy.

Materials and Methods

Source Population
The Institutional Review Board of the Mayo Foundation approved the research protocol. A comprehensive Mayo
Clinic centralized diagnostic index, utilizing inpatient and
outpatient discharge diagnoses, pathology reports, and endoscopic reports, identified all patients with CUC evaluated at
the Mayo Clinic Rochester between January 1, 1976, and
December 31, 2002 (N approximately 11,800). Inclusion
into the study as either a case or control required a clinical,
histologic, and endoscopic or radiographic diagnosis of CUC
made or confirmed at the Mayo Clinic and availability of full
case note documentation. At least 1 study author confirmed
the diagnosis of CUC for all cases and controls, assessing study
eligibility without knowledge of risk variable data. Exclusion
of patients with an unclear diagnosis of CUC, indeterminate
colitis, or features suggestive of Crohns disease occurred prior
to data collection.

Case and Control Identication

Cases and controls, selected from the same source
population described above, composed the study population.
The same comprehensive index identified a total of 297 potential cases with colorectal adenocarcinoma from the source
population. Selection as a case required a clinical and histologic
diagnosis of colorectal adenocarcinoma made at the Mayo
Clinic and a CRC diagnosis at least 2 years after the onset of
CUC. At least 1 study author confirmed the diagnosis of CRC,
assessing eligibility criteria for the study without knowledge
of risk variable data. Potential cases were excluded because of
the following: incomplete medical records (n 28); historical
diagnosis of CRC (n 28); unconfirmed diagnosis of CUC (n

GASTROENTEROLOGY Vol. 130, No. 7

22); coincident diagnosis of CUC and CRC (n 13);

diagnosis of CRC within 2 years of the diagnosis of CUC (n
6); diagnosis of a colonic neoplasm other than adenocarcinoma
(n 5); and CUC diagnosed after CRC (n 4). Suitable
controls could not be found for 3 cases. A total of 188 cases
were matched to controls.
Not all CUC-specific matching information is part of the
centralized index; therefore, controls were first selected based
on gender and date of their first Mayo Clinic visit for colitis.
A total of 1528 potential controls (8:1 ratio of potential
controls:case) were selected, mirroring the distribution of cases
for these 2 variables. Abstraction of these charts, without
knowledge of case data, provided the remaining matching
variables.
To be eligible as a control, the duration of colitis had to
match or exceed the cases duration of colitis. Cases and
controls were randomly matched in a 1:1 ratio based on 5
criteria: gender, maximal extent of CUC, duration of CUC,
date of first Mayo Clinic visit for CUC (within 3 years), and
calendar year of CUC diagnosis (within 5 years). The last 3
criteria ensured comparison of similar calendar periods between cases and controls as well as similar follow-up at the
Mayo Clinic compared with elsewhere. We categorized the
extent of CUC as follows: proctitis (up to 15 cm from the anal
verge); proctosigmoiditis (up to 35 cm); left-sided colitis (up
to 55 cm or distal to the splenic flexure); and extensive colitis
(greater than 55 cm or proximal to the splenic flexure).

Data Collection
In this retrospective study, cases were followed from
the date of diagnosis of CUC until the development of cancer,
whereas controls were followed from the date of diagnosis until
a date that matched the duration of colitis in the corresponding case. Exposures or events occurring during this period were
recorded uniformly for cases and controls; those occurring
afterward (eg, 5-ASA use or diagnosis of PSC) were not recorded. Review of all inpatient and outpatient notes; surgery,
pathology, radiology, and endoscopy reports; and patient encounter forms provided a comprehensive history for data abstraction. Patient encounter forms, completed by all patients at
each clinic visit, provided systematic and up-to-date personal
information about family history of CRC, patient tobacco use
(current use, start and quit dates), current prescription and
nonprescription medications, and medication allergies. All
information on this form is reviewed by the treating physician
for completeness and accuracy at each patient visit.

Denition of Variables
We recorded patient, clinical, endoscopic, and therapeutic factors identified in the literature as associated or potentially associated with CRC risk among patients with CUC.
We also collected additional demographic information and
data on potential confounders. All data were registered on a
standardized form using prespecified definitions of variables.
Demographic information abstracted includes the following: gender; ethnicity; age at CUC diagnosis; age at CRC

June 2006

diagnosis (cases) or equivalent follow-up (controls); and geographic residence, categorized as either Minnesota, 5-state
region (Iowa, Illinois, North Dakota, South Dakota, and Wisconsin), or elsewhere. Patient variables collected include the
following: diagnosis of CRC in a first-degree relative; tobacco
use at CUC diagnosis (current smoker, exsmoker, or never
smoker); continuous tobacco use for more than 1 year after a
diagnosis of CUC; and documented or self-reported allergy to
sulfa with evidence of no further use of sulfasalazine.
Clinical and endoscopic variables recorded include the following: radiographically or histologically confirmed PSC; 1
average physician contacts per year; 2 exacerbations per year
(an exacerbation was defined as a medication change or increase
because of symptoms); presence of terminal ileal inflammation
(ie, backwash ileitis); number of barium enemas performed
after the diagnosis of CUC; number of prior surveillance
colonoscopies (1, 1 or 2, 2); and a diagnosis of pseudopolyps made at any previous colonic evaluation. We defined a
surveillance examination as a colonoscopic procedure to the
cecum with multiple random biopsy specimens taken at regular intervals throughout the colon and performed for the
purpose of detecting dysplasia and not in the presence of active
symptoms or in response to a change in symptoms.
Treatment variables recorded include the following: cumulative duration of use of oral 5-ASA agents, oral prednisone,
and immunosuppressive agents (azathioprine, 6-mercaptopurine). All formulations of oral mesalamine and 5-ASA prodrugs
(sulfasalazine, olsalazine, balsalazide) were recorded. Duration
of use was categorized as 1 year, 15 years, 6 10 years, and
10 years for 5-ASA and 1 year for other medications. Use
of nonprescription medications and supplements, including
folic acid, multivitamin, calcium, aspirin, and nonsteroidal
anti-inflammatory drugs (NSAIDs), noted on at least 2 office
visits was also recorded. No patients reported use of selective
cyclooxygenase-2 (COX-2) antagonists. To prevent misclassification, data on ursodeoxycholic acid (UDCA) use for PSC
patients were not reported because a significant proportion of
PSC patients participated in long-term clinical trials of
UDCA, and unblinded treatment data were unavailable.

Statistical Analysis
The association between individual variables and CRC
was assessed using conditional logistic regression to account
for the matching of cases and controls. These analyses examined the association for each of the individual variables of
interest adjusting for duration of CUC diagnosis and age.
These analyses are labeled as univariate analyses throughout
the remainder of this text. The set of primary variables of
interest was then examined in a multiple variable model
analysis using a backward elimination approach to identify a
final set of variables independently associated with CRC; age
and duration of CUC were included automatically in this
analysis, a cut-off level of 0.1 was used to retain variables in
the backward elimination approach. Although not statistically
significant, PSC was included in the final model as a potential
confounder between use of anti-inflammatory therapy and

COLORECTAL CANCER PREVENTION IN UC

1943

Table 1. Demographics of Patients With CUC-Associated

CRC and CUC Patients Without CRC

Characteristic
Male sex, n (%)a
Extensive colitis, n (%)a
Median duration of CUC, y
(range)a
White, n (%)b
Median age at CUC diagnosis, y
(range)
Median age at cancer diagnosis
or follow-up, y (range)
Geographic residence, n (%)
Minnesota
Five-state area
Elsewhere

CUC-associated
CRC (cases)
(n 188)

CUC patients
without CRC
(controls)
(n 188)

133 (71)
159 (85)
17 (351)

133 (71)
159 (85)
17 (352)

186 (99)
25 (676)

187 (99)
27 (866)

45 (1784)

46 (1683)

56 (30)
68 (36)
64 (34)

39 (21)
79 (42)
70 (37)

CUC, chronic ulcerative colitis; CRC, colorectal cancer.

aSex, extent of colitis, and duration of CUC were used to match cases
and controls.
bTwo cases were Hispanic; 1 control was African American.

CRC and between performance of surveillance colonoscopy and

CRC. Immunosuppressive use was included to permit a complete assessment of all anti-inflammatory therapy. Inclusion of
PSC and immunosuppressive use in the model did not significantly change the results. All results are presented as odds
ratios (OR) with the associated 95% confidence intervals (95%
CI). Those confidence intervals not containing unity were
considered statistically significant.

Results
Study Population
Controls were gender- and disease extent-matched
(Table 1). They were chosen to be similar to matched
cases with respect to the duration of CUC. The majority
of the study population was male and white. Most patients had extensive colitis; only 2 cases and 2 controls
(1%) had proctitis. The distribution of geographic residence was approximately equal among cases and controls
and evenly spread across the 3 categories. Approximately
half of the cases were diagnosed before the midpoint of
the study in 1989, and half were diagnosed after (95 vs
93, respectively). All CRCs occurred within an area of
active or quiescent colitis.
Univariate Patient Factors
A diagnosis of CRC in a first-degree relative was
the only patient factor significantly associated with CRC
(Table 2). No other variable demonstrated a significant
association, but cigarette smoking was somewhat protective. Sulfa allergy was not associated with CRC.

1944

VELAYOS ET AL

GASTROENTEROLOGY Vol. 130, No. 7

Table 2. Univariate Analysis of Potential Patient Factors Associated With CRC in CUC
Characteristic

Cases (n 188)

Controls (n 188)

ORa

95% CI

Family history CRC, n (%)

Smoking status at CUC diagnosis, n (%)
Nonsmoker
Exsmoker
Current
Smoker after CUC diagnosis, n (%)
Sulfa allergy, n (%)

18 (10)

6 (3)

3.0b

1.27.7

141 (75)
20 (11)
27 (14)
37 (20)
18 (10)

124 (66)
26 (14)
38 (20)
51 (27)
21 (11)

1.0
0.5
0.6
0.7
1.0

0.21.2
0.31.1
0.41.1
0.51.8

CUC, chronic ulcerative colitis; CRC, colorectal cancer; OR, odds ratio; 95% CI, 95% confidence interval.
aOdds ratio from a conditional logistic regression model including the matching factors and the individual characteristic listed.
bP .05.

Univariate Clinical and Endoscopic Factors

A prior diagnosis of pseudopolyps was significantly associated with CRC (Table 3). Surveillance
colonoscopy was associated with an incremental stepwise
protection against CRC. A diagnosis of PSC, number of
physician contacts, frequent exacerbations of colitis,
backwash ileitis, and performance of barium enema were
not significantly associated with CRC.
Univariate Treatment Factors
No treatment variables were found to be significantly associated with CRC, although a trend between
immunosuppressive therapy for 1 year and CRC was
observed (Table 4). All such patients received azathioprine, except for 1 control treated with 6-mercaptopurine. All patients except for 1 case and 2 controls received
less than the currently recommended dose for maintenance therapy of 2.0 2.5 mg/kg daily for azathioprine or
1.0 1.5 mg/kg daily for 6-mercaptopurine. For these
patients, the average dose of azathioprine was 1.0 mg/kg
daily for cases and 1.3 mg/kg daily for controls.
Corticosteroid use for 1 year and 5-ASA use for 1
year had a protective association with CRC; however, the
latter was only statistically significant for the 15 year
duration. A total of 217 patients used sulfasalazine for

more than 1 year (101 cases, 116 controls), 32 used

mesalamine (8 cases, 24 controls), 2 used olsalazine (2
controls), and none used balsalazide. A total of 17 patients used both sulfasalazine and mesalamine for more
than 1 year (4 cases, 13 controls).
There was a strong and significant protective effect
observed with over-the-counter use of aspirin and
NSAIDs. Although duration of use was not quantified,
most patients were on aspirin or NSAIDs for a clear
chronic indication including arthritis, arthralgias, spondylitis, spondyloarthropathy, chronic headache, chronic
gout, chronic low back pain, or cardiac prophylaxis.
Among cases, 7 aspirin users (78%) and 3 NSAIDs users
(60%) had such an indication, whereas, among controls,
26 aspirin users (84%) and 17 NSAIDs users (81%) had
such an indication. Among users of either aspirin or
NSAIDs, 10 of 14 cases (71%) and 30 of 45 controls
(67%) were over 50 years of age. Protective trends observed with consumption of supplemental calcium, multivitamin, or folate failed to reach statistical significance.
Multiple Variable Model
The backward elimination conditional logistic
analysis, adjusted for age at CUC diagnosis and the
duration of CUC, identified the most influential variables

Table 3. Univariate Analysis of Potential Clinical and Endoscopic Factors Associated With CRC in CUC
Characteristic

Cases (n 188)

Controls (n 188)

ORa

95% CI

Primary sclerosing cholangitis, n (%)

1 physician contact/y, n (%)
2 exacerbations/y, n (%)
Backwash ileitis, n (%)
1 barium enema after diagnosis, n (%)
Surveillance colonoscopy, n (%)
1
12
2
Pseudopolyps, n (%)

26 (14)
81 (43)
9 (5)
40 (21)
68 (36)

24 (13)
95 (51)
12 (6)
28 (15)
63 (33)

1.1
0.8
0.7
1.5
0.9

0.62.1
0.51.1
0.31.7
0.92.5
0.61.6

133 (70)
39 (21)
16 (9)
105 (56)

91 (48)
60 (32)
37 (20)
79 (42)

1.0
0.5b
0.2b
2.0b

0.30.8
0.10.5
1.32.1

CUC, chronic ulcerative colitis; CRC, colorectal cancer; OR, odds ratio; 95% CI, 95% confidence interval.
ratio from a conditional logistic regression model including the matching factors and the individual characteristic listed.
bP .05.
aOdds

June 2006

COLORECTAL CANCER PREVENTION IN UC

1945

Table 4. Univariate Analysis of Potential Therapeutic Factors Associated With CRC in CUC
Characteristic
IBD medications
Duration of 5-ASA therapy, n (%)
1 y
15 y
610 y
10 y
Corticosteroid use 1 y, n (%)
Immunosuppressive therapy 1 y, n (%)
Non-IBD treatment, n (%)
Aspirin
NSAIDs
Calcium
Multivitamin
Folate

Cases (n 188)

Controls (n 188)

ORa

95% CI

83 (44)
38 (20)
34 (18)
33 (18)
39 (21)
12 (6)

60 (32)
47 (25)
39 (21)
42 (22)
48 (26)
7 (4)

1.0
0.5b
0.6
0.6
0.6b
2.1

0.30.9
0.31.1
0.31.0
0.40.9
0.77.2

9 (5)
5 (3)
9 (5)
21 (11)
16 (9)

31 (17)
21 (11)
14 (7)
30 (16)
17 (9)

0.2b
0.2b
0.7
0.6
0.9

0.10.5
0.10.6
0.31.5
0.41.3
0.41.8

CUC, chronic ulcerative colitis; CRC, colorectal cancer; n, number; OR, odds ratio; 95% CI, 95% confidence interval; IBD, inflammatory bowel
disease; NSAIDs, nonsteroidal antiinflammatory drugs.
aOdds ratio from a conditional logistic regression model including the matching factors and the individual characteristic listed.
bP .05.

independently associated with CRC in the study population (Table 5). A prior diagnosis of pseudopolyps remained significantly associated with CRC, even after
adjusting for surveillance colonoscopy and use of antiinflammatory therapy. Family history of CRC appeared
to be important as well (P .05). Smoking for at least
1 year after the diagnosis of CUC, more than 1 surveilTable 5. Final Multiple Variable Model of Significant
Predictive and Protective Factors Independently
Associated With CRC in CUC
Variables
Patient Variables
Family history of CRC
Smoking after CUC diagnosis
Clinical and endoscopic variables
Primary sclerosing cholangitis
Surveillance colonoscopy
1
1 or 2
2
Pseudopolyps
Treatment variables
IBD therapy
5-ASA use
1 y
15 y
610 y
10 y
Corticosteroid use 1 y
Immunosuppressive use 1 y
Non-IBD Therapy
Aspirin
NSAIDs

ORa

95% CI

3.7b
0.5c

1.013.2
0.20.9

1.1

0.52.3

1.0
0.4c
0.3c
2.5c

0.20.7
0.10.8
1.44.6

1.0
0.4c
0.6
0.6
0.4c
3.0

0.20.9
0.31.4
0.31.3
0.20.8
0.713.6

0.3c
0.1c

0.10.8
0.030.5

aOdds ratio based on a backward elimination conditional logistic

regression model.
bP .05.
cP .05.

lance colonoscopy, 5-ASA therapy for 15 years, corticosteroid use of more than 1 year, and aspirin and
NSAID use all demonstrated a significant protective
effect. Longer term 5-ASA use demonstrated a protective
trend that was not statistically significant. Inclusion of
PSC and immunosuppressive use in the final model did
not significantly change the results. Our final model was
the same whether we adjusted or not for geographic
residence.

Discussion
To identify potentially important CRC prevention strategies and high-risk subgroups, we investigated
several predictive and protective factors associated with
CRC. In a large tertiary care population of CUC patients,
the 2 most important predictive factors for CRC (after
matching for anatomic extent and duration of disease)
were having a first-degree relative with CRC and having
a prior diagnosis of pseudopolyps. The most influential
protective factors (again after matching for anatomic
extent and duration of disease) were surveillance colonoscopy, smoking, and use of anti-inflammatory medications including 5-ASA agents, corticosteroids, aspirin,
and NSAIDs. In a multivariate model, surveillance
colonoscopy reduced cancer odds by 60% and 5-ASA
therapy by approximately 50%, although the latter was
no longer statistically significant after 5 years. After
adjusting for surveillance colonoscopy and use of antiinflammatory medications, a prior diagnosis of pseudopolyps remained a significant cancer risk factor.
We confirmed a recent St. Marks Hospital study
reporting an interesting association between postinflammatory pseudopolyps and CRC.23 Rutter et al23 reported

1946

VELAYOS ET AL

a 2.3-fold increase in odds of CRC for pseudopolyps

diagnosed on surveillance colonoscopy, only slightly
lower than the 2.5-fold increase in odds we observed for
pseudopolyps diagnosed on barium enema or colonoscopy. Pseudopolyps were common in both studies. A
total of 62% of cases and 39% of controls had postinflammatory pseudopolyps in the St. Marks study, compared with 56% of cases and 42% of controls in our
study.
Pseudopolyps are not inherently premalignant.24 They
are irregular islands of mucosa that form because of acute
inflammation and regeneration. In our study, biopsied
polypoid lesions proven to be neoplastic were not categorized as pseudopolyps. Therefore, the association between pseudopolyps and CRC may represent either improper recognition of malignant polypoid lesions,
mucosal lesions obscured by pseudopolyps, or difference
in the number of biopsy specimens taken in patients with
and without pseudopolyps or that pseudopolyps are a
visible surrogate marker of significant inflammation predisposing to cancer. Given the high frequency of pseudopolyps in the control patients (42%), prophylactic
colectomy is not likely to be an attractive management
strategy for patients with this common and often challenging endoscopic finding. However, a more intensive
surveillance strategy could be considered.
Of particular interest is the protective effect we observed with both surveillance colonoscopy and several
anti-inflammatory medications. Prior studies of one of
these factors have not adjusted for the mutual effect of
the other. Barium enema did not afford protection
against CRC; however, at least 1 surveillance colonoscopy reduced cancer risk by 62%, similar to the 57%
reduction in CRC mortality reported in a populationbased study.15
Surveillance colonoscopy likely protects against CRC
via the performance of colectomy after dysplasia is detected and before cancer sets in. It often does not involve
an endoscopic procedure (eg, polypectomy) that directly
reduces CRC risk. Our study was not designed to answer
this point. To study the association between various
medications and cancer risk, controls required an intact
or partially intact colon. Therefore, controls clearly were
not protected from CRC because of proctocolectomy after
the detection of dysplasia. It may therefore be that our
results more accurately describe that a lack of surveillance is a risk factor for CRC, rather than ongoing
surveillance being a protective factor. Alternatively, the
protective effect we observed with surveillance may be
due to other protective factors that we did not record,
including duration from previous colonoscopy, dose of
anti-inflammatory medication, endoscopic interventions,

GASTROENTEROLOGY Vol. 130, No. 7

or partial colectomy. Adjustment for frequency of exacerbations, physician visits, type of inflammatory bowel
disease (IBD) medication, geographic residence, years of
clinic follow-up, and restriction of our analysis to the
highest risk patients with more than 10 years of disease
did not change our summary estimate. Overall, our
findings are consistent with a recent Cochrane review
concluding that, even among high-quality studies, the
evidence in support of surveillance colonoscopy reducing
the risk of CRC is present, although indirect.25
It is noteworthy that 55 patients developed CRC
despite undergoing a prior surveillance examination.
Physicians have been advised to counsel patients on the
limitations of this procedure for over a decade.16 We did
not register variables associated with the sensitivity of
the test, including the number of biopsies,26 frequency of
surveillance, patient compliance, prior dysplasia, and a
newly reported variable, procedure duration.27 Despite
demonstrating a protective effect compared with no
screening, our study highlights that strategies for improving the sensitivity and efficacy of surveillance
colonoscopy are needed.
Our study is the largest case-control study to date
examining the role of 5-ASA therapy on CRC risk,
permitting simultaneous evaluation and adjustment for
multiple variables of interest. Our study is novel in that
we analyzed whether the risk of cancer associated with
5-ASA therapy was independent of surveillance colonoscopy and tested a broader array of prescription and
over-the-counter anti-inflammatory medications. We
also framed our epidemiologic study within a meaningful
clinical context to assess whether anti-inflammatory therapy reduces the risk of CRC in CUC.
We found a lower risk of CRC associated with smoking and a variety of anti-inflammatory therapies, supporting the hypothesis that reduction of chronic inflammation can alter the progression from inflammation to
cancer. Smoking28 and transdermal nicotine29,30 have
been shown to alleviate the clinical course of IBD, presumably through reduction of inflammation. The antiinflammatory properties of 5-ASA drugs, corticosteroids,
aspirin, and NSAIDs are well known, even though the
exact molecular mechanism by which reduction of inflammation may decrease cancer risk is not. 5-ASA is
structurally similar to aspirin and NSAIDs and shares
similar molecular targets, interfering with inflammation,
proliferation, and apoptosis.31
Smoking, corticosteroids, and other anti-inflammatory
therapy have been reported to reduce the risk of colitisrelated CRC in several observational studies.4,6 Use of
aspirin or NSAIDs has been reported to reduce colorectal
neoplasia by 40%50% in non-IBD populations,32,33 and

June 2006

NSAIDs have been reported to reduce CRC mortality

odds by 49% in a population of United States military
veterans with IBD.34 A recent meta-analysis of 9 observational studies reported that 5-ASA use reduced the
odds of colitis-related CRC by 49%.35 After adjusting for
surveillance colonoscopy and other anti-inflammatory
therapy, we observed a 50% reduction in cancer odds
associated with 15 years of 5-ASA use, comparable with
2, large, population-based studies reporting that 5-ASA
reduced cancer odds between 49% and 61%.14,22
Arguing against the hypothesis that reduction of
chronic inflammation prevents CUC-related CRC is that
immunosuppressive use did not reduce cancer odds.
However, only 1 case and 2 controls were taking the
recommended weight-adjusted dose of azathioprine for
suppressing inflammation in long-term maintenance. A
recent study also could not demonstrate a reduction in
neoplasia risk with immunosuppressive use36; however,
another study observed a nonsignificant reduction in
CRC risk.18 The reason for the observed difference in
cancer odds between immunosuppressive and other antiinflammatory therapy requires further study. It may be
that differences in the molecular properties of these
medications, the reduction of inflammation achieved
with each medication, or the neoplasia risk of the underlying population taking each medication may be relevant
factors.
Information on long-term use of over-the-counter
medications and supplements was not available for all
patients. We therefore inferred causality for our NSAIDs
and aspirin data based on most patients using these
over-the-counter medications for a chronic indication.
We also could not calculate a dose response for 5-ASA
use because information on dose was not available for all
patients. A dose response was not seen with duration of
use. Several case-control studies6,22,26 have reported that
a dose of 1.2 g of mesalamine or its equivalent and
regularity of 5-ASA use are associated with reduced CRC
risk. Perhaps these combined variables, and not duration
of use, may be better measures for assessing a dose
response.
We did not address the effect of anti-inflammatory
medication on dysplasia risk because our question of
interest related to cancer prevention. However, clearly,
more studies are needed to demonstrate a causal link
between anti-inflammatory therapy preventing cancer
through the prevention of dysplasia.35
Similar to previous studies, medication history was
obtained from chart review. This raises the possibility of
exposure misclassification, particularly for over-thecounter medications and supplements such as calcium
and vitamins. We attempted to minimize this possibility

COLORECTAL CANCER PREVENTION IN UC

1947

by abstracting medication data from standardized forms

completed by all patients prior to each office visit that
have start and stop dates, are confirmed by the treating
physician, and have been an integral part of the permanent chart for decades. We confirmed that there was not
a difference in reporting of various medication exposures
between local and more distant patients; even so, we
acknowledge that a true protective association among
calcium, vitamins, or folate could have been masked.
We addressed potential referral bias and secular trends
by adjusting for location of residence, years of clinic
follow-up, and matching on year of first clinic visit. Our
primary exposure estimates for 5-ASA use, NSAIDs use,
and surveillance colonoscopy are comparable with those
reported in large population-based studies in which the
concern for referral bias should not exist; even so, we
acknowledge that our study population may not be fully
representative of CUC patients seen primarily outside of
a tertiary care setting. Therefore, valid inference for our
study should be limited to our intended target population: patients with CUC cared for in a tertiary care
center.
Unexpectedly, we did not find that PSC was a risk
factor for cancer. However, our results are consistent with
a prior cohort13 and case-control study12 performed at the
Mayo Clinic. One potential explanation is that the Mayo
Clinic serves as a referral center for investigational therapy for PSC and has conducted clinical trials with UDCA
for this indication. Many, if not most, patients with PSC
cared for at the Mayo Clinic have received therapy with
UDCA, which is known to have a chemopreventive effect
for colonic dysplasia.37,38
Our results, combined with compatible biologic data,
suggest that chronic use of anti-inflammatory therapy
may be a rational and promising cancer prevention strategy in CUC. Additional studies are needed to test further
this hypothesis and to strengthen the causal chain of
evidence between pharmacologic therapy reducing biologic surrogates of inflammation or neoplasia and then in
turn reducing dysplasia and cancer. These studies may
also provide insight on how to optimize this strategy
clinically. Until more data are available, we prescribe
maintenance 5-ASA or immunosuppressive medication
(depending on the clinical indication) at the appropriate
dose, with the primary goal of maintaining clinical remission. A secondary benefit of this practice may be the
prevention of CRC. Because many questions remain, we
follow the recommendations of a recently published consensus guideline39 and do not alter surveillance practices
based on the use of maintenance anti-inflammatory
therapy.

1948

VELAYOS ET AL

In conclusion, to identify potentially important CRC

prevention strategies and high-risk subgroups, we investigated several predictive and protective factors associated with CRC within a large tertiary care population of
CUC patients. Although several questions remain to be
answered, these results suggest that, in a population
matched for extent and duration of CUC, surveillance
colonoscopy and use of anti-inflammatory medications
may reduce the risk of CRC. A history of postinflammatory pseudopolyps appears to be a predictive factor for
cancer.

References
1. Itzkowitz SH, Yio X. Inflammation and cancer IV. Colorectal cancer
in inflammatory bowel disease: the role of inflammation. Am J
Physiol Gastrointest Liver Physiol 2004;287:G717.
2. Ekbom A, Helmick C, Zack M, Adami HO. Ulcerative colitis and
colorectal cancer. A population-based study. N Engl J Med 1990;
323:1228 1233.
3. Prior P, Gyde SN, Macartney JC, Thompson H, Waterhouse JA,
Allan RN. Cancer morbidity in ulcerative colitis. Gut 1982;23:
490 497.
4. Lashner BA, Heidenreich PA, Su GL, Kane SV, Hanauer SB. Effect
of folate supplementation on the incidence of dysplasia and
cancer in chronic ulcerative colitis. A case-control study. Gastroenterology 1989;97:255259.
5. Nuako KW, Ahlquist DA, Mahoney DW, Schaid DJ, Siems DM,
Lindor NM. Familial predisposition for colorectal cancer in chronic
ulcerative colitis: a case-control study. Gastroenterology 1998;
115:1079 1083.
6. Eaden J, Abrams K, Ekbom A, Jackson E, Mayberry J. Colorectal
cancer prevention in ulcerative colitis: a case-control study. Aliment Pharmacol Ther 2000;14:145153.
7. Askling J, Dickman PW, Karlen P, Brostrom O, Lapidus A, Lofberg
R, Ekbom A. Family history as a risk factor for colorectal cancer in
inflammatory bowel disease. Gastroenterology 2001;120:1356
1362.
8. Kornfeld D, Ekbom A, Ihre T. Is there an excess risk for colorectal
cancer in patients with ulcerative colitis and concomitant primary
sclerosing cholangitis? A population based study. Gut 1997;41:
522525.
9. Shetty K, Rybicki L, Brzezinski A, Carey WD, Lashner BA. The risk
for cancer or dysplasia in ulcerative colitis patients with primary
sclerosing cholangitis. Am J Gastroenterol 1999;94:16431649.
10. Heuschen UA, Hinz U, Allemeyer EH, Stern J, Lucas M, Autschbach F, Herfarth C, Heuschen G. Backwash ileitis is strongly
associated with colorectal carcinoma in ulcerative colitis. Gastroenterology 2001;120:841 847.
11. Greenstein AJ, Sachar DB, Smith H, Pucillo A, Papatestas AE,
Kreel I, Geller SA, Janowitz HD, Aufses AH Jr. Cancer in universal
and left-sided ulcerative colitis: factors determining risk. Gastroenterology 1979;77:290 294.
12. Nuako KW, Ahlquist DA, Sandborn WJ, Mahoney DW, Siems DM,
Zinsmeister AR. Primary sclerosing cholangitis and colorectal
carcinoma in patients with chronic ulcerative colitis: a casecontrol study. Cancer 1998;82:822 826.
13. Loftus EV Jr, Sandborn WJ, Tremaine WJ, Mahoney DW, Zinsmeister AR, Offord KP, Melton LJ III. Risk of colorectal neoplasia in
patients with primary sclerosing cholangitis. Gastroenterology
1996;110:432 440.
14. Pinczowski D, Ekbom A, Baron J, Yuen J, Adami HO. Risk factors
for colorectal cancer in patients with ulcerative colitis: a casecontrol study. Gastroenterology 1994;107:117120.

GASTROENTEROLOGY Vol. 130, No. 7

15. Karlen P, Kornfeld D, Brostrom O, Lofberg R, Persson PG, Ekbom

A. Is colonoscopic surveillance reducing colorectal cancer mortality in ulcerative colitis? A population based case control study.
Gut 1998;42:711714.
16. Bernstein CN, Shanahan F, Weinstein WM. Are we telling patients
the truth about surveillance colonoscopy in ulcerative colitis?
Lancet 1994;343:7174.
17. Bernstein CN, Blanchard JF, Metge C, Yogendran M. Does the
use of 5-aminosalicylates in inflammatory bowel disease prevent
the development of colorectal cancer? Am J Gastroenterol 2003;
98:2784 2788.
18. Rutter M, Saunders B, Wilkinson K, Rumbles S, Schofield G,
Kamm M, Williams C, Price A, Talbot I, Forbes A. Severity of
inflammation is a risk factor for colorectal neoplasia in ulcerative
colitis. Gastroenterology 2004;126:451 459.
19. Bernstein CN. Challenges in designing a randomized trial of
surveillance colonoscopy in IBD. Inflamm Bowel Dis 1998;4:
132141.
20. Rubin DT, Lashner BA. Will a 5-ASA a day keep the cancer (and
dysplasia) away? Am J Gastroenterol 2005;100:1354 1356.
21. Moody GA, Jayanthi V, Probert CS, Mac Kay H, Mayberry JF.
Long-term therapy with sulphasalazine protects against colorectal cancer in ulcerative colitis: a retrospective study of colorectal
cancer risk and compliance with treatment in Leicestershire. Eur
J Gastroenterol Hepatol 1996;8:1179 1183.
22. van Staa TP, Card TR, Leufkens HG, Logan RF. 5-Aminosalicylate
use and colorectal cancer risk in inflammatory bowel disease: a
large epidemiological study. Gut 2005;54:15731578.
23. Rutter MD, Saunders BP, Wilkinson KH, Rumbles S, Schofield G,
Kamm MA, Williams CB, Price AB, Talbot IC, Forbes A. Cancer
surveillance in longstanding ulcerative colitis: endoscopic appearances help predict cancer risk. Gut 2004;53:18131816.
24. Kelly JK, Gabos S. The pathogenesis of inflammatory polyps. Dis
Colon Rectum 1987;30:251254.
25. Mpofu C, Watson AJ, Rhodes JM. Strategies for detecting colon
cancer and/or dysplasia in patients with inflammatory bowel
disease. Cochrane Database Syst Rev 2004:CD000279.
26. Rubin CE, Haggitt RC, Burmer GC, Brentnall TA, Stevens AC, Levine
DS, Dean PJ, Kimmey M, Perera DR, Rabinovitch PS. DNA aneuploidy in colonic biopsies predicts future development of dysplasia
in ulcerative colitis. Gastroenterology 1992;103:16111620.
27. Toruner M, Harewood GC, Loftus EV Jr, Sandborn WJ, Tremaine
WJ, Faubion WA, Schroeder KW, Egan LJ. Endoscopic factors in
the diagnosis of colorectal dysplasia in chronic inflammatory
bowel disease. Inflamm Bowel Dis 2005;11:428 434.
28. Logan RF, Edmond M, Somerville KW, Langman MJ. Smoking and
ulcerative colitis. Br Med J (Clin Res Ed) 1984;288:751753.
29. Sandborn WJ, Tremaine WJ, Offord KP, Lawson GM, Petersen BT,
Batts KP, Croghan IT, Dale LC, Schroeder DR, Hurt RD. Transdermal nicotine for mildly to moderately active ulcerative colitis. A
randomized, double-blind, placebo-controlled trial. Ann Intern
Med 1997;126:364 371.
30. Pullan RD, Rhodes J, Ganesh S, Mani V, Morris JS, Williams GT,
Newcombe RG, Russell MA, Feyerabend C, Thomas GA, et al.
Transdermal nicotine for active ulcerative colitis. N Engl J Med
1994;330:811 815.
31. Allgayer H. Review article: mechanisms of action of mesalazine in
preventing colorectal carcinoma in inflammatory bowel disease.
Aliment Pharmacol Ther 2003;18(Suppl 2):10 14.
32. Thun MJ, Namboodiri MM, Heath CW Jr. Aspirin use and reduced
risk of fatal colon cancer. N Engl J Med 1991;325:15931596.
33. Smalley W, Ray WA, Daugherty J, Griffin MR. Use of nonsteroidal
anti-inflammatory drugs and incidence of colorectal cancer: a
population-based study. Arch Intern Med 1999;159:161166.
34. Bansal P, Sonnenberg A. Risk factors of colorectal cancer in
inflammatory bowel disease. Am J Gastroenterol 1996;91:44
48.

June 2006

35. Velayos FS, Terdiman JP, Walsh JM. Effect of 5-aminosalicylate

use on colorectal cancer and dysplasia risk: a systematic review
and metaanalysis of observational studies. Am J Gastroenterol
2005;100:13451353.
36. Matula S, Croog V, Itzkowitz S, Harpaz N, Bodian C, Hossain S,
Ullman T. Chemoprevention of colorectal neoplasia in ulcerative
colitis: the effect of 6-mercaptopurine. Clin Gastroenterol Hepatol 2005;3:10151021.
37. Tung BY, Emond MJ, Haggitt RC, Bronner MP, Kimmey MB, Kowdley
KV, Brentnall TA. Ursodiol use is associated with lower prevalence of
colonic neoplasia in patients with ulcerative colitis and primary sclerosing cholangitis. Ann Intern Med 2001;134:8995.
38. Pardi DS, Loftus EV Jr, Kremers WK, Keach J, Lindor KD. Ursodeoxycholic acid as a chemopreventive agent in patients with ulcerative colitis and primary sclerosing cholangitis. Gastroenterology
2003;124:889 893.

COLORECTAL CANCER PREVENTION IN UC

1949

39. Itzkowitz SH, Present DH. Consensus conference: colorectal cancer screening and surveillance in inflammatory bowel disease.
Inflamm Bowel Dis 2005;11:314 321.

Received September 8, 2005. Accepted March 1, 2006.

Address requests for reprints to: Edward V. Loftus Jr, MD, Division of
Gastroenterology and Hepatology, Mayo Clinic, 200 First Street Southwest, Rochester, Minnesota 55905. e-mail: loftus.edward@mayo.edu;
fax: (507) 266-0335.
Supported in part by Procter and Gamble Pharmaceuticals and the
Mayo Foundation for Medical Education and Research.
Presented in part at the 106th Annual Meeting of the American
Gastroenterological Association, May 14 19, 2005, Chicago, IL (Gastroenterology 2005;128(4 Suppl 2):A328).

Houston of the Rectal Valves of Houston

John Houston (18021845) was born in the north of Ireland and, at an early age, was adopted by a physician
uncle. An avid interest in anatomy led him to become curator of the museum at the Royal College of Surgeons
in Dublin. In 1826, he was awarded an MD degree by Edinburgh University. Returning to Dublin, he was
appointed to the surgical staff of the newly established municipal hospital where he established his reputation
as a keen clinical observer and a deft operator. In 1830, he published a description of the prominent mucosal
folds that characterize the lumen of the rectum. His untimely demise came when he was fatally stricken by a
cerebral hemorrhage while giving a lecture to students.
Contributed by WILLIAM S. HAUBRICH, MD
The Scripps Clinic, La Jolla, California