Aliment Pharmacol Ther 2000; 14: 145-153.
`
`Colorectal cancer prevention in ulcerative colitis:
`a case-control study
`
`J. EADEN*. K. ABRAMSt, A. EKBOMt, E. JACKSON* & J. MAYBERRY*
`*The Gastrointestinal Research Unit, Leicester General Hospital, Leicester, UK; tDepartment of Epidemiology and Public
`Health, University of Leicester, Leicester, UK; and tDepartment of Medical Epidemiology, Karolinska Institute, Stockholm,
`Sweden
`
`Accepted for publication 4 October 1999
`
`SUMMARY
`
`Background: The risk of colorectal cancer (CRC) in
`ulcerative colitis (UC) increases with extent and dura-
`tion of disease. Identifying other risk factors would allow
`targeting of sub-groups at greatest risk, enabling more
`cost-effective surveillance.
`Methods: We conducted a case-control study comparing
`102 cases of CRC in UC with matched controls. Odds
`ratios (OR) for cancer risk were estimated by conditional
`logistic regression. A multivariate model assessed the
`contribution of individual variables.
`Results: Regular 5-aminosalicylic acid (5-ASA) therapy
`reduces cancer risk by 75% (OR 0.25, 95% CI: 0.13-
`
`0.48, P < 0.00001). Adjusting for other variables,
`taking mesalazine regularly reduces risk by 81% (OR
`0.19, 95% CI: 0.06-0.61, P = 0.006) and visiting a
`hospital doctor more than twice a year also reduces risk
`(OR 0.16, 95% CI: 0.04-0.60, P = 0.007). Considering
`variables independently, having a family history of
`sporadic CRC in any relative increases risk fivefold (OR
`5.0, 95% CI: 1.10-22.82, P < 0.04).
`Conclusions: CRC risk among UC patients can be reduced
`by regular therapy with 5-ASA medication. Colon-
`oscopic surveillance may be best targeted on those
`unable to take 5-ASAs (e.g. due to allergy) and those
`with a positive family history of CRC.
`
`INTRODUCTION
`
`Colorectal cancer is one of the most serious complica-
`tions of ulcerative colitis (IJC). The risk becomes
`significant after 8-10 years of colitis and increases at
`a rate of 0.5-1% between the second and fourth decades
`of disease.' After 40 years of pancolitis approximately
`25-30% of patients will have developed colorectal
`cancer.2 The risk of colorectal cancer is not related to
`duration of disease alone but also to its extent."
`However, the severity and frequency of attacks do not
`confer an increased risk.5' 6 There is some evidence that
`if the onset of UC is at a young age, the risk of malignant
`transformation is increased independent of either dis-
`
`Correspondence to: Dr I. Eaden. The Gastrointestinal Research Unit.
`Leicester General Hospital. Gwendolen Road. Leicester LE5 4PW. UK.
`E-mail: jayne.a.eaden@btinternet.com
`
`ease duration or anatomic extent, although this is
`disputed.'-' 7-9
`Several studies have suggested that if patients with UC
`also have primary sclerosing cholangitis they may be at
`a higher risk of developing colorectal cancer.1°-12 The
`evidence for other potential risk factors is scarce.
`A positive family history of colon cancer,13' 14 smok-
`ing15 and folate depletion'4' 16 may affect the occur-
`rence of colorectal cancer. Aspirin is thought to have an
`anti-neoplastic effect in the large bowel" and regular
`consumption of low-dose aspirin reduces the risk of
`adenomatous polyps and fatal colon cancer in the
`general population.18' 19 There is growing evidence that
`the chronic consumption of aminosalicylates, in partic-
`ular sulphasalazine. may also provide some protection
`against colorectal cancer in patients with ulcerative
`colitis through a similar mechanism of action.15. 20-22
`
`© 2000 Blackwell Science Ltd
`
`145
`
`SALIXN00011404
`
`Dr. Falk Ex. 2017
`GeneriCo v. Dr. Falk IPR2016-00297
`Page 1
`
`

`
`146 (cid:9)
`
`i. EADEN et al.
`
`As these studies are few in number we wished to
`investigate this hypothesis further whilst also studying
`the effect of other 5-ASA compounds that have
`previously been neglected.
`The optimal study design for defining risk factors
`would be a prospective, controlled study. However,
`given the long time needed for sufficient cancers or
`dysplasias to develop in a surveillance programme, as
`well as ethical concerns about withholding surveillance
`colonoscopy or 5-ASA medications from the control
`group, we must still rely on case-control studies to offer
`the best approximation of colorectal cancer risk factors
`in ulcerative colitis. With this in mind we aimed to
`assess the risk factors thought to play a part in the
`development of colorectal cancer in UC and to build a
`statistical model that would identify the most hazardous
`combination of factors.
`
`MATERIALS AND METHODS
`
`The investigation was designed as a retrospective
`matched case-control study. In order to identify cases,
`one author (J.E.) contacted consultant gastroenterolo-
`gists across England and Wales and asked for permis-
`sion to review the medical records of their patients with
`known colorectal cancer complicating UC. Nineteen
`gastroenterologists (eight from teaching hospitals and
`11 from district general hospitals) who were interested
`in the study agreed to a search of their patient records
`and/or pathology databases. Once potential cases had
`been identified the same author visited each hospital
`and systematically recorded various details from each
`patient's record on a pro forma.
`
`Cases and controls
`
`For subjects to be included in the study the diagnosis of
`UC had to be confirmed clinically, histologically and
`radiologically. The criteria used were those established
`by Lennard-Jones.23 Cases who were deceased at the
`time of the study were included provided the medical
`notes had not been destroyed. Cases were excluded if
`they had been referred with a diagnosis of CRC where
`UC was an incidental finding and if full case note
`documentation was not available.
`From 133 cases collected. 102 met the inclusion
`criteria and these were matched with controls from
`the Leicestershire inflammatory bowel disease patient
`database, which was rigorously assembled during the
`
`late 1980s using established international diagnostic
`criteria.24 The matching criteria were (i) sex; (ii) age
`within 10 years; (iii) extent of disease at the time of
`diagnosis of UC; and (iv) duration of disease within a
`5-year window. In addition controls had to have an
`intact colon and not have a colorectal cancer at the
`time of diagnosis of the case. It was not possible to
`match cases with a control from the same hospital as
`most hospitals do not have a database of their IBD
`patients.
`
`Data collection
`
`Information was extracted from all in-patient and out-
`patient medical records for each subject from the date of
`diagnosis of UC until the date of the patient's cancer
`diagnosis. The data for cases were extracted and
`recorded in the same manner by one author (J.E.).
`Data from control notes were independently extracted
`by two authors (J.E and E.J.) as a quality control check
`that data retrieval was uniform and accurate. All
`medical notes (whether or not they pertained to UC)
`were reviewed so that a comprehensive history, in
`particular family history, could be obtained. Data
`extracted included:
`
`(a) Age at diagnosis of UC.
`(b) Pharmacotherapy: treatment for the 5-10 years
`prior to the development of cancer including 5-ASA
`preparations, corticosteroids (systemic and local)
`and aspirin. If there was a significant period of time
`1 year) during which a subject was not taking
`medication, either because it had been stopped by a
`doctor or if a subject was documented as being a
`poor complier with medication, they were recorded
`as not taking regular medication.
`(c) Average frequency of contacts with a hospital
`physician or surgeon per year over the course of
`their disease.
`(d) Number of barium enemas and colonoscopic exam-
`inations during follow-up of their UC.
`(e) Activity of UC: each subject was placed into one of
`six categories as follows: (i) silent disease; (ii) one
`exacerbation every 10 years; (iii) one exacerbation
`per 1-10 years; (iv) one exacerbation per month to
`1 year; (v) one exacerbation per month; and
`(vi) continuous symptoms. In the final calculations
`this variable was reduced to three categories (see
`Table 2) to prevent small sample sizes in the
`statistical analysis.
`
`© 2000 Blackwell Science Ltd. Aliment Pharmacol Ther 14, 145-153
`
`SALIXN00011405
`
`Dr. Falk Ex. 2017
`GeneriCo v. Dr. Falk IPR2016-00297
`Page 2
`
`

`
`COLORECTAL CANCER PREVENTION IN ULCERATIVE COLITIS 147
`
`(1) Smoking history at the time of diagnosis of UC.
`(g) Presence of primary sclerosing cholangitis (PSC)
`confirmed at endoscopic retrograde cholangio-pan-
`creatography/percutaneous transhepatic cholangio-
`graphy/liver biopsy. Serological values of raised
`liver function tests for more that 1 year (in the
`absence of PSC) were also recorded.
`(h) Positive family history of IBD and colorectal cancer.
`
`For cases, the age at cancer diagnosis and its site and
`stage were also recorded. We would have liked to
`investigate the effect of folate on colorectal cancer risk.
`After examining 20 sets of notes it was obvious that
`folate levels were not routinely measured/recorded in
`the medical notes and therefore this variable was not
`studied further.
`
`Statistical analysis
`
`The study was designed to have a power of 80% to
`detect an odds ratio of 2.5 at the level of 5% significance
`assuming a prevalence of 65% for each risk factor in the
`control group. Conditional logistic regression was used
`to compute estimates of odds ratio (OR) as a measure of
`association between various exposures and colorectal
`cancer, together with 95% confidence intervals. Risk
`parameters b-f were analysed as categorical variables in
`the final analysis. Model development used changes in
`minus twice the log-likelihood to assess the contribution
`of individual variables in a forward selection procedure,
`with variables being added to the model if the change
`was statistically significant at the 5% level.
`
`RESULTS
`
`The characteristics of cases and controls are summar-
`ized in Table 1. The mean age at the time of diagnosis of
`colorectal cancer was 57.4 years (s.d. ± 12.9) and the
`mean interval between diagnosis of UC and colorectal
`cancer was 16.1 years (s.d. ± 9.7). For controls the
`mean duration of disease was 16 years (s.d. ± 9.4).
`Fifty-seven per cent of cancers were located in the
`rectosigmoid with the remainder evenly distributed
`around the rest of the colon. Typically, over the course
`of their disease, cases were much less likely to take
`medication on a regular basis, had fewer contacts with
`their hospital physician and had fewer colonoscopic
`examinations. Subjects with cancer had more family
`members with a history of sporadic colorectal cancer
`but the activity of disease, number of barium enemas
`
`© 2000 Blackwell Science Ltd. Aliment Pharmarol Ther 14, 145-153
`
`and family history of IBD did not differ significantly
`between cases and controls. The number of subjects
`with primary sclerosing cholangitis and raised serol-
`ogical liver function tests was small and therefore could
`not be investigated further. Of the 51 cases who took a
`5-ASA compound on a regular basis, 37 took
`sulphasalazine (six patients < 2 g/day. 31 patients
`2 g/day), 12 took mesalazine (one patient < 1.2 g/
`day, 11 patients 1.2 g/day) and two took other drugs.
`In comparison, of the 84 controls receiving a 5-ASA
`compound, 39 took sulphasalazine (seven patients
`< 2 g/day, 32 patients > 2 g/day), 43 took mesalazine
`(five patients < 1.2 g/day, 38 patients 1.2 g/day) and
`two took other drugs.
`The independent effect of each variable on the odds
`ratio of developing colonic cancer is shown in Table 2.
`The most significant finding was the strong protective
`association of regular 5-ASA therapy, reducing
`cancer risk by 75% (OR 0.25, 95% CI: 0.13-0.48,
`P< 0.00001). When individual 5-ASA drugs and their
`doses were analysed, mesalazine at a dose of 1.2 g/day
`or greater reduced colorectal cancer risk by 91%
`compared to no treatment (OR 0.09, 95% CI: 0.03-
`0.28, P< 0.00001) and was also protective when
`taken at lower doses (OR 0.08, 95% CI: 0.08-0.85,
`P = 0.04). The benefits of sulphasalazine were less
`pronounced and an effect was only evident for a dose of
`2 g/day or greater (OR 0.41, 95% CI: 0.18-0.92,
`P= 0.03). Other 5-ASA medications had a non-signi-
`ficant protective effect. Frequent visits to see a hospital
`doctor were also highly protective (OR 0.098, 95% CI:
`0.03-0.29, P< 0.00001), as was having between one
`and two colonoscopies over the history of their colitis
`(OR 0.22, 95% CI: 0.09-0.55, P< 0.001). Systemic
`and local steroid therapy also have a statistically
`significant protective role but a dose-response effect
`was not demonstrated for either route of administration.
`Smoking history, particularly being an ex-smoker at the
`time UC was diagnosed, was associated with a non-
`significant protective effect. A positive family history of
`colorectal cancer in any family member increased
`cancer risk by a factor of five (OR 5.00, 95% CI:
`1.10-22.82, P< 0.04), but when only first-degree
`relatives were considered this fell to 3.5 and was no
`longer significant (OR 3.50, 95% CI: 0.73-16.85,
`P= 0.11). Aspirin use had a minimal protective role
`which was not statistically significant, but this may be
`due to the small numbers taking this therapy in our
`study.
`
`SALIXN00011406
`
`Dr. Falk Ex. 2017
`GeneriCo v. Dr. Falk IPR2016-00297
`Page 3
`
`

`
`148 (cid:9)
`
`1. EADEN et al.
`
`Sex
`Males
`Females
`
`Ethnic origin
`Caucasian
`Asian
`
`Age at diagnosis of UC
`<15
`15-29
`30-49
`50+
`
`Cases [n (%)]
`
`Controls [n (%)]
`
`Table 1. Characteristics of cases and con-
`trols
`
`64 (63)
`38 (37)
`
`96 (94)
`6 (6)
`
`3(3)
`27 (27)
`36 (35)
`36 (35)
`
`64 (63)
`38 (37)
`
`58 (57)
`44 (43)
`
`1(1)
`22 (22)
`42 (41)
`37 (36)
`
`Mean age at diagnosis of UC (s.d.)
`Mean duration of disease (s.d.)
`
`41.33 (± 16.8)
`16.1 (± 9.7)
`
`43.59 (± 16.1)
`16 (± 9.4)
`
`Extent at diagnosis
`Proctitis
`Left sided
`SubtotaUtotal
`
`Physician/surgeon contacts per year
`< 1
`1-2
`2+
`
`Number of colonoscopies (over UC history)
`< 1
`1-2
`2+
`
`Number of barium enemas (over UC history)
`< 1
`1-2
`2+
`
`Activity of UC
`Silent disease
`1 exacerbation/10 years
`1 exacerbation/1-10 years
`More frequent
`
`Smoking history at diagnosis of UC
`Never smoked
`Current smoker
`Ex-smoker
`
`Primary sclerosing cholangitis
`Raised liver function tests (no PSC)
`
`Family history of colorectal cancer
`Family history of [BD
`
`Pharmacotherapy
`Regular use of 5-ASA preparation
`Regular use of systemic steroid
`Regular use of local steroid
`Regular use of aspirin
`
`6 (6)
`34 (33)
`62 (61)
`
`32 (31)
`59 (58)
`11 (11)
`
`26 (26)
`42 (41)
`34 (33)
`
`19 (19)
`57 (56)
`26 (26)
`
`22 (22)
`38 (37)
`31 (30)
`11 (11)
`
`76 (74)
`11 (11)
`15 (15)
`
`1 (1)
`12 (12)
`
`10 (10)
`5 (5)
`
`51 (50)
`5 (5)
`8 (8)
`4 (4)
`
`5 (5)
`33 (32)
`64 (63)
`
`9 (9)
`58 (57)
`35 (34)
`
`8 (8)
`64 (63)
`30 (29)
`
`26 (26)
`56 (55)
`20 (20)
`
`20 (20)
`41 (40)
`30 (29)
`11 (11)
`
`68 (67)
`12 (12)
`22 (22)
`
`0
`11 (10.8)
`
`2 (2)
`5 (5)
`
`84 (82)
`19 (19)
`18 (18)
`5 (5)
`
`© 2000 Blackwell Science Ltd. Aliment Pharmacol Ther 14, 145-153
`
`SALIXN00011407
`
`Dr. Falk Ex. 2017
`GeneriCo v. Dr. Falk IPR2016-00297
`Page 4
`
`

`
`Table 2. Independent effect of characteristics on colorectal cancer risk
`
`COLORECTAL CANCER PREVENTION IN ULCERATIVE COLITIS 149
`
`Odds ratio 95% CI
`
`P-value
`
`Smoked at time UC diagnosed (compared to non-smoker) 0.71
`Ex-smoker when UC diagnosed (compared to non-smoker) 0.53
`No
`Yes
`Mesalazine (< 1.2 g/day)
`Mesalazine (cid:9)
`1.2 g/day)
`Sulphasalazine (< 2 g/day)
`Sulphasalazine (cid:9)
`2 g/day)
`Other (e.g. olsalazide, balsalazide)
`Yes (compared to none)
`Yes (compared to none)
`Yes (compared to none)
`1 to 2 visits per year (compared to none)
`More than 2 visits per year (compared to none)
`Any (compared to none)
`1 to 2 over course of disease (compared to none)
`More than 2 over course of disease (compared to none)
`
`0.28-1.8
`0.23-1.22
`
`0.47
`0.14
`
`0.13-0.48 < 0.00001
`0.08-0.85
`0.04
`0.03-0.28 < 0.00001
`0.17-1.84
`0.34
`0.18-0.92
`0.03
`0.04-3.58
`0.41
`0.01-0.70
`0.008
`0.19-1.02
`0.06
`0.21-2.98
`0.74
`0.14-0.76
`0.009
`0.03-0.29 < 0.00001
`0.88-1.29
`0.50
`0.09-0.55
`0.001
`0.16-1.10
`0.08
`0.41-3.15
`0.80
`1.10-22.82
`0.04
`0.73-16.85
`0.11
`0.40-1.82
`0.67
`0.35-2.60
`0.92
`0.34-2.60
`0.90
`
`0.25
`0.08
`0.09
`0.56
`0.41
`0.40
`0.26
`0.44
`0.80
`0.32
`0.098
`1.07
`0.22
`0.42
`1.14
`5.00
`3.50
`0.85
`0.95
`0.93
`
`Variable
`
`Smoking
`
`5-ASA
`
`Systematic steroid
`Local steroid
`Aspirin
`Contact with hospital doctor
`
`Barium enema
`Colonoscopy
`
`Raised alk phos (no PSC)
`Family history of colorectal cancer Any relative
`1st degree relative
`1 exacerbation/10 years
`1 exacerbation/1-10 years
`1 exacerbation/month-1 year (or more frequent)
`
`Activity of disease
`
`Table 3. Adjusted odds ratios for most influential variables
`
`Variable
`
`5-ASA
`
`Contact with hospital doctor
`
`CRC in any relative
`
`Colonoscopies after diagnosis
`
`After adjustment for individuals 5-ASA drugs
`Mesalazine
`
`Sulphasalazine
`
`Other
`Contact with hospital doctor
`
`CRC in any relative
`
`Colonoscopies after diagnosis
`
`Odds ratio 95% CI
`
`P-value
`
`None
`Yes
`<1
`1 to 2 per year over the course of disease
`> 2 per year over the course of disease
`No
`Yes
`<1
`1 to 2 over the course of disease
`> 2 over the course of disease
`None
`< 1.2 g/day
`1.2 g/day
`< 2 g/day
`a 2 g/day
`Variable doses
`< 1
`1 to 2 per year over the course of disease
`> 2 per year over the course of disease
`No
`Yes
`< 1
`1 to 2 over the course of disease
`> 2 over the course of disease
`
`0.47
`
`0.43
`0.19
`
`6.38
`
`0.27
`0.52
`
`0.18
`0.19
`0.93
`0.85
`1.21
`
`0.42
`0.16
`
`6.84
`
`0.33
`0.55
`
`0.22-1.00
`
`0.05
`
`0.16-1.15
`0.06-0.65
`
`0.09
`0.008
`
`0.97-41.96
`
`0.05
`
`0.09-0.77
`0.17-1.56
`
`0.02
`0.24
`
`0.02-1.92
`0.06-0.61
`0.22-3.91
`0.32-2.26
`0.08-18.97
`
`0.15-1.18
`0.04-0.60
`
`0.16
`0.006
`0.92
`0.75
`0.89
`
`0.10
`0.007
`
`0.80-58.60
`
`0.08
`
`0.11-1.01
`0.18-1.71
`
`0.05
`0.30
`
`© 2000 Blackwell Science Ltd. Aliment Pliarinneol Timer 14, 145-153
`
`SALIXN00011408
`
`Dr. Falk Ex. 2017
`GeneriCo v. Dr. Falk IPR2016-00297
`Page 5
`
`

`
`150 (cid:9)
`
`i. EADEN et al.
`
`A suitable model was developed to assess the contri-
`bution of the variables in a forward selection procedure.
`The final model included regular 5-ASA therapy,
`frequent contacts with a hospital doctor, a positive
`family history of sporadic colonic cancer in any relative
`and one to two colonoscopies over the course of UC
`history. Table 3 shows the effect of these variables, in
`terms of odds ratios, adjusted for the other variables in
`the model. Regular consumption of mesalazine at a dose
`of 1.2 g/day or greater (OR 0.19, 95% CI: 0.06-0.61,
`P= 0.006) and frequent visits to a hospital physician
`(OR 0.16, 95% CI: 0.04-0.60, P= 0.007) confer the
`greatest benefit after adjusting for the other variables.
`We investigated the possibility of all interactions
`between the variables in the final model and none
`was statistically significant at the 5% level.
`As all the controls in this study came from one area
`(Leicester) we carried out a further analysis stratified
`by case to assess whether this may have biased the
`results. There were 12 case-control pairs in which
`both the case and control were from Leicester. These
`data were compared with the other 90 pairs where the
`control came from Leicester but the case came from
`elsewhere. Regarding 5-ASA medication the indepen-
`dent OR for non-Leicester pairs was 0.25 (95% CI:
`0.13-0.51, P= 0.0001) and for Leicester pairs was
`0.2 (95% CI: 0.02-1.71, P= 0.14). It is interesting to
`note that the point estimate is therefore lower for the
`Leicester pairs, with the analysis only losing its
`statistical significance due to a small numbers effect.
`Looking specifically at mesalazine 1.2 g/day, the OR
`for non-Leicester pairs was 0.14 (95% CI: 0.05-0.41,
`P= 0.0003) and for Leicester pairs was 0.25 (95% CI:
`0.03-2.24, P= 0.2). The effect is not as marked in the
`Leicester pairs but 5-ASA is still protective (again the
`statistical significance is lost due to a small numbers
`effect).
`There were 44 Asian control patients compared to six
`Asian cases. We therefore conducted a stratified analysis
`based on ethnicity to assess if this may have biased the
`results. There were 57 pairs where both the case and
`control were Caucasian and an analysis of the indepen-
`dent effect of 5-ASA medication on these pairs gave an
`OR of 0.30 (95% CI: 0.13-0.65, P= 0.003). Regular
`consumption of mesalazine at a dose of 1.2 g/day was
`also still highly protective with an OR of 0.1 (95% CI:
`0.02-0.39, P= 0.001). Having 1-2 colonoscopies over
`the course of their disease was protective (OR = 0.16,
`95% CI: 0.05-0.55, P= 0.003) as were frequent visits
`
`to a hospital doctor (OR = 0.17. 95% CI: 0.04-0.65.
`P= 0.01).
`Finally we performed a stratified analysis to assess if
`there was any difference in the results from teaching vs.
`district hospitals. There were 29 pairs where the case and
`control came from teaching hospitals and 73 pairs where
`the case alone was from a district hospital. 5-ASAs were
`protective in both groups with identical odds ratios of
`0.25. The effect of mesalazine (cid:9)
`1.2 g/day) also gave
`similar results with an OR of 0.1 in teaching hospitals
`(95% CI: 0.01-0.78, P= 0.03) and 0.18 in district
`hospitals (95% CI: 0.06-0.53, P= 0.02).
`
`DISCUSSION
`Colorectal cancer (CRC) risk in patients with UC can be
`substantially reduced by taking 5-ASA therapy on a
`regular basis. After adjusting for other variables,
`mesalazine is particularly effective, reducing the cancer
`risk by 81%. This protective effect is independent of dose
`but becomes statistically significant for 1.2 g/day or
`greater. The benefits of sulphasalazine are not as
`pronounced and a significant effect is only seen at
`higher doses (cid:9)
`2 g/day). These protective effects are
`independent of disease activity. A family history of
`sporadic colorectal cancer in any relative is associated
`with a fivefold increased risk, although this did not quite
`reach statistical significance after adjustment for other
`variables in the analysis. Systemic and local steroid use
`also have an inverse relationship with the development
`of colorectal cancer although their effects were not as
`influential in the model analysis.
`Previous studies15. 20. 21 have suggested a protective
`role for sulphasalazine against the development of
`colorectal cancer and recent research has shown that
`mesalazine selectively induces apoptosis of tumour cells
`in sporadic colorectal cancer.22 Our study not only agrees
`with these findings, but has also allowed an analysis of
`the effects of other 5-ASA compounds, and has demon-
`strated that mesalazine exerts an even greater protective
`effect than sulphasalazine. It is postulated that NSAIDs
`and 5-ASA compounds work in a similar way and may
`reduce CRC risk by inhibiting mucosal prostaglandin
`synthesis.25 This is supported by three lines of research.
`Firstly, there is a reduced risk of large bowel adenomas
`among aspirin and NSAID users.17. 18. 26 Secondly,
`NSAIDs decrease the number and size of colorectal
`adenomas in patients with familial adenomatous polypo-
`s 7 and the morbidity and mortality rates for CRC are
`
`© 2000 Blackwell Science Ltd. Aliment Pharmneol Ther 14, 145-153
`
`SALIXN00011409
`
`Dr. Falk Ex. 2017
`GeneriCo v. Dr. Falk IPR2016-00297
`Page 6
`
`

`
`COLORECTAL CANCER PREVENTION IN ULCERATIVE COLITIS 151
`
`low in patients on chronic NSAIDs.18. 25. 28 Finally,
`NSAIDs have been shown to decrease the number and
`size of chemically induced colon adenomas and carcino-
`mas in experimental animal studies.29' 3° In our study
`only 3.9% of cases and 4.9% of controls were taking
`aspirin. These small numbers limit any interpretation of
`its role. However, it does mean that the beneficial effects of
`5-ASA compounds are not due to the co-incidental use of
`aspirin.
`Patients who see a hospital doctor of any grade
`frequently and have at least one colonoscopy are less
`likely to develop bowel cancer. However, these actions
`do not necessarily reduce the risk of malignancy per se
`and probably represent markers for compliance. This is
`also likely to be associated with a high compliance
`taking prescribed medication. Thus, a protective effect
`found for pharmacological agents may be an under-
`estimation, as the controls are more likely to be
`compliant even in this respect compared with cases.
`As controls appear to be more compliant than cases,
`one may wonder whether this was because the controls
`were aware that they were part of the Leicestershire
`database and thus were more motivated or health
`conscious than the cases, who were followed in a
`greater variety of practice settings without a cohesive
`database/registry. However, we feel that this is unlikely
`as the database is used entirely for epidemiological
`evaluations and is not used for clinical follow-up.
`A higher colonoscopy rate among the controls rein-
`forces our results because it makes it unlikely that this
`group will have a high frequency of undetected cancers.
`Although colonoscopy has not been proven to be of
`beneficial effect with regard to reducing the colorectal
`mortality in this patient group, it is a consistent finding
`in most studies that being subjected to a surveillance
`programme upgrades the Duke stage when the cancers
`are diagnosed.31. 32 In some reported studies of surveil-
`lance, cancers are still missed even with 6 monthly
`examinations and patients should understand that a
`colonoscopy is not an absolute guarantee against
`malignant transformation.33' 34
`A positive family history of CRC is an established risk
`factor for the disease in the general population35 and in
`our study having any family member with a sporadic
`colonic cancer increased the risk for patients with UC by
`a factor of five. The relationship was not maintained
`during the development of a model, and became non-
`significant when we analysed first-degree relatives only,
`but this may be a small numbers effect. Indeed research
`
`© 2000 Blackwell Science Ltd. Aliment Phartnneol Ther 14, 145-153
`
`from the Mayo clinic has shown that a family history of
`colorectal cancer in first-degree relatives was twice as
`common in UC patients with CRC than in UC controls
`matched for extent and duration of colitis.13
`The strengths of our study include a uniform approach
`to data retrieval and the retrieval of all medical notes for
`both cases and controls. A second investigator independ-
`dently extracted data from control notes using the same
`pro forma. as a quality control check that the information
`had been retrieved accurately. A national database of IBD
`patients does not exist in the UK, and thus case
`identification could not have been carried out by any
`other method than the one chosen. Controls were
`identified from our local database and thus in some
`instances differed in geographical location and ethnicity
`from the cases. We do not feel this had a bearing on our
`results. We have shown that the odds ratio for 5-ASA
`medication is actually lower in the 12 pairs from Leicester
`compared with the other 90 pairs, the analysis only losing
`its statistical significance due to a small numbers effect.
`Furthermore when we analysed the data after removing
`Asian cases and controls from the investigation we found
`that 5-ASA medication (including mesalazine), frequent
`visits to a hospital doctor and 1-2 colonoscopies were all
`still highly protective against colorectal cancer in the
`colitis population. In addition, there have been no
`documented studies stating that colorectal cancer risk
`differs in a non-Caucasian population or varies with
`geographical location across the UK. Indeed Kochhar's
`study36 from India stated that the crude incidence of CRC
`in Asian UC patients was comparable to the 3-4%
`incidence reported from Anglo-Saxon countries.
`A factor that is crucial to consider in studies of cancer
`risk in UC is the colectomy rate in the population
`studied. If the colectomy rate was higher in Leicester
`compared to other geographic areas, that would
`eliminate patients from the pool at risk for developing
`CRC and leave behind only very low-risk individuals as
`controls. Probert's study37 demonstrates a similar
`colectomy rate for patients in Leicestershire (in partic-
`ular Asians) compared with reported rates for
`St Marks38 and North-East Scotland39 and thus this
`should not be a source of bias in our study.
`Criticisms could be levelled at the methods of deter-
`mining the length and dosage of pharmacological
`therapy continuously over long periods of time. This is
`a valid criticism because doses differ with duration of
`disease and therefore an average dose was estimated for
`each subject over the course of their colitis. If there was
`
`SALIXN00011410
`
`Dr. Falk Ex. 2017
`GeneriCo v. Dr. Falk IPR2016-00297
`Page 7
`
`

`
`152 (cid:9)
`
`J. EADEN et al.
`
`any doubt whatsoever as to whether a subject was
`taking medication (or if they had their treatment
`discontinued for longer than 1 year) they were classed
`as non-compliers and so our findings are based on
`subjects clearly identified as taking medication on a
`regular and virtually continuous basis. Patients who
`had their medication temporarily interrupted (e.g. a
`pregnant woman who feared adverse drug effects to the
`foetus) were recorded as being compliant. All retrospec-
`tive studies that rely on retrieving information from
`medical records are subject to some inaccuracies (e.g. in
`medication usage) and it could be suggested that direct
`patient interview would be more suitable. However, this
`is not possible in our study as some cases and controls
`were deceased. Also, patient recall of facts is fraught
`with inaccuracies and would introduce another source
`of bias. We feel that the medical record is the most
`legitimate source of data as it provides a contemporary
`record made at the time of consultation.
`The strong associations that we found for regular
`5-ASA therapy and frequent visits to a hospital
`physician and the risk of developing colorectal cancer
`are likely to have an important impact on care
`programmes and screening for colorectal cancer in
`patients suffering from ulcerative colitis. The cost-
`effectiveness of a colonoscopic surveillance programme
`in ulcerative colitis has been questioned by many
`authors 40-42 Physicians may choose to better target
`colonoscopic surveillance on those who are at greatest
`risk, i.e. those unable to take regular 5-ASAs (for
`example due to allergy), have a positive family history of
`CRC and perhaps have primary sclerosing cholangitis.
`Likewise it should be possible to reduce the frequency of
`examinations in those at lower risk. We believe that this
`effort and expense may be better directed at educating
`and encouraging patients to take their medication
`regularly. Such an approach should be supported by
`hospital-based follow-up for all patients' although
`patients have to take a modicum of responsibility for
`their illness. The combination of seeing a hospital doctor
`on a regular basis, compliance with medication and
`attendance at colonoscopy offers the greatest degree of
`protection against colorectal cancer that patients can
`control themselves.
`
`ACKNOWLEDGEMENTS
`
`We would like to thank the following consultant
`gastroenterologists across the UK who kindly gave their
`
`permission for access to medical patient records: Dr A.
`Catterall, Dr I. M. Chesner, Dr M. W. Dronfield, Dr J. A.
`Gibson, Mrs C. Hall, Dr A. B. Hawthorne, Dr K. Kane,
`Dr S. Kane, Dr A. J. Lobo, Dr R. G. Long, Mr B. V.
`Palmer, Prof. J. Rhodes, Prof. J. M. Rhodes, Dr H.
`Shepherd, Dr P. Smith, Dr R. H. Teague, Dr H. H. Tsai,
`Dr A. J. Turnbull and Dr P. J. Winwood.
`
`REFERENCES
`
`1 Ransohoff DF. Colon cancer in ulcerative colitis. Gastroen-
`terology 1988; 94: 1089-91.
`2 Ekbom A, Helmick C, Zack M, Adami HO. Ulcerative colitis
`and colorectal cancer. A population-based study. N Engl J Med
`1990; 323: 1228-33.
`3 Bargen JA, Gage RP. Carcinoma and ulcerative colitis: prog-
`nosis. Gastroenterology 1960; 39: 385-93.
`4 Gyde SN, Prior P, Allan RN, et al. Colorectal cancer in ulcer-
`ative colitis: a cohort study of primary referrals from three
`centres. Gut 1988; 29: 206-17.
`5 Katzka I, Brody RS, Morris E, Katz S. Assessment of colorectal
`cancer risk in patients with ulcerative colitis: experience from
`a private practice. Gastroenterology 1983; 85: 22-9.
`6. Lennard-Jones JE, Morson BC, Ritchie JK, Williams CB. Cancer
`surveillance in ulcerative co