Severity of inflammation is a risk factor for colorectal neoplasia in ulcerative colitis

GASTROENTEROLOGY 2004;126:451– 459

Severity of Inflammation Is a Risk Factor for Colorectal Neoplasia in Ulcerative Colitis MATTHEW RUTTER,*,‡ BRIAN SAUNDERS,‡ KAY WILKINSON,‡ STEVE RUMBLES,‡ GILLIAN SCHOFIELD,‡ MICHAEL KAMM,‡ CHRISTOPHER WILLIAMS,‡ ASHLEY PRICE,‡ IAN TALBOT,‡ and ALASTAIR FORBES‡ *Department of Gastroenterology, University Hospital of North Tees, Stockton-on-Tees, Teesside, England; and ‡St. Mark’s Hospital, Harrow, London, England

Background & Aims: Patients with ulcerative colitis are at increased risk of colorectal cancer. It is widely believed that this is secondary to colonic inflammation. However, the severity of colonic inflammation has never been shown to be a risk factor. Methods: We devised a case-control study of patients with long-standing extensive ulcerative colitis to examine various potential risk factors for neoplasia. All cases of colorectal neoplasia detected from our surveillance program between January 1, 1988, and January 1, 2002, were studied (n ⴝ 68). Each patient was matched with 2 control patients from the same surveillance population (n ⴝ 136). Matching was for sex, colitis extent, age at onset, duration of colitis, and year of index surveillance colonoscopy. Segmental colonoscopic and histological inflammation was recorded by using a simple score (0, normal; 1, quiescent/chronic inflammation; and 2, 3, and 4, mild, moderate, and severe active inflammation, respectively). Other data collected included history of primary sclerosing cholangitis, family history of colorectal cancer, and smoking and drug history (mesalamine 5-aminosalicylic acid, azathioprine, and folate). Results: Univariate analysis showed a highly significant correlation between the colonoscopic (odds ratio, 2.5; P ⴝ 0.001) and histological (odds ratio, 5.1; P < 0.001) inflammation scores and the risk of colorectal neoplasia. No other factors reached statistical significance. On multivariate analysis, only the histological inflammation score remained significant (odds ratio, 4.7; P < 0.001). Conclusions: In long-standing extensive ulcerative colitis, the severity of colonic inflammation is an important determinant of the risk of colorectal neoplasia. Endoscopic and histological grading of inflammation could allow better risk stratification for surveillance programs.

atients with ulcerative colitis are at increased risk of colorectal cancer, but within this population there is considerable variability in the cancer risk. Although it is possible that the variability is genetic, with an individual co-inheriting the risk of developing both ulcerative colitis and colorectal cancer,1 the most widely held belief is

P

that the increased cancer risk is primarily acquired, probably secondary to colonic inflammation.2,3 Supporting evidence includes the fact that duration of colitis is a risk factor for colorectal cancer. However, the severity of colonic inflammation has never been shown to be a risk factor for colorectal cancer. Current colonoscopic surveillance programs for cancer in ulcerative colitis focus on patients with long-standing, extensive disease but often do not refine the surveillance protocol further. Although many centers, including our own, believe that these programs are worthwhile and national guidelines from both the United States and the United Kingdom advocate surveillance, there is little doubt that surveillance is both labor intensive and expensive. Further refinement of high- and low-risk subgroups of patients is highly desirable because it would allow a more appropriate allocation of resources, reducing unnecessary procedures for those at lower risk of developing colorectal cancer while enabling more intensive surveillance for those patients most likely to benefit. We devised a case-control study of patients with longstanding extensive ulcerative colitis to examine various potential risk factors for colorectal neoplasia (cancer or dysplasia), with particular focus on the severity of colonic inflammation, smoking history, and prescribed drug use.

Materials and Methods Surveillance Program St. Mark’s Hospital has a well-established cancer surveillance program for patients with long-standing, extensive ulcerative colitis. Patients with histologically proven ulcerative colitis who have macroscopic inflammation extending proximal to the splenic flexure are counseled about their Abbreviation used in this paper: PSC, primary sclerosing cholangitis. © 2004 by the American Gastroenterological Association 0016-5085/04/$30.00 doi:10.1053/j.gastro.2003.11.010

452

RUTTER ET AL.

increased risk of colorectal cancer and are offered colonoscopic surveillance every 2 years from 8 years after the onset of colitic symptoms. During each colonoscopy, the segmental macroscopic appearance is documented on a computerized database (Metabase; Metasa Ltd., London, UK), and segmental mucosal biopsy samples are taken. Patients also undergo clinical assessment in the intervening year, at which time rigid sigmoidoscopy and mucosal biopsy is performed. Any biopsy samples showing dysplasia are independently reported by 2 experienced pathologists, and when dysplasia is confirmed, colectomy is advised.

GASTROENTEROLOGY Vol. 126, No. 2

2.

Cases and Controls A case-control study was devised to investigate putative risk factors for colorectal neoplasia (dysplasia or cancer) in ulcerative colitis. From a total population of 525 patients who had undergone colonoscopic surveillance between January 1, 1988, and January 1, 2002, all cases of colorectal neoplasia were studied. Each patient was matched with 2 control patients. The control population was derived from our prospective ulcerative colitis surveillance database. Cases and controls were matched for sex and extent of ulcerative colitis. Patients were also matched for age at the onset of colitic symptoms and duration of ulcerative colitis (to within 5 years). These 2 matches thus ensured that the ages of cases and controls were also matched. In addition, control patients had to be on surveillance with an intact colon to within 5 years of the time of diagnosis of their respective case’s neoplasia. To ensure that patients of similar eras were being compared, patients were also matched for the year of index surveillance colonoscopy (to within 5 years).

3.

4.

5.

Data Collection Data collected included colonoscopic and histological grade of inflammation, history of primary sclerosing cholangitis (PSC), family history of colorectal cancer, and smoking history as contemporaneously recorded. Data on mesalamine, azathioprine, and folate supplement use were also recorded. Information was collected from our prospective surveillance database, patient case notes, and colonoscopy and histopathology reports. For patients still under surveillance, this information was supplemented by patient interviews and postal questionnaires. More detailed information on data collection and categorization is as follows: 1. Colonoscopic grade of inflammation—this was determined by use of a simple 5-point score: 0, entirely normal appearance; 1, quiescent disease (mild edema or chronic features, but no active inflammation); 2, mild active inflammation; 3, moderate active inflammation; and 4, severe active inflammation. Our colonoscopy report system requires segmental descriptions (rectum, sigmoid, descending, splenic flexure, transverse, hepatic flexure, ascending colon, and caecum). From the information provided on the colonoscopy report, each seg-

6.

7.

8.

ment of the colon was designated a score, and a mean score for that colonoscopy was derived. The mean score of all surveillance colonoscopies for that patient was used to assign each patient an overall colonoscopic inflammation grade. Histological grade of inflammation—a similar method was used to determine the histological grade of inflammation by using a scoring system based on epithelial neutrophils: 0, normal (no inflammatory cells); 1, chronic inflammation only; 2, mild active (cryptitis but no crypt abscesses); 3, moderate active (few crypt abscesses); and 4, severe active inflammation (numerous crypt abscesses). Segmental colonic mucosal biopsy samples were taken at each colonoscopy. When more than 1 biopsy sample had been taken from a segment of colon, the highest grade of histological inflammation within that segment was recorded. The mean histological score for all segments of each colonoscopy was calculated, and the mean histological score of all surveillance colonoscopies for that patient was used to assign each patient an overall histological inflammation grade. PSC—patients were considered positive for PSC only if the diagnosis had been radiologically or histologically confirmed. Smoking history—patients were categorized as current smokers, ex-smokers, or nonsmokers. For a separate analysis, patients were recategorized into those who had smoked since the onset of their colitis and those who had either stopped smoking before the onset of colitis or had never smoked. Family history of colorectal cancer—patients were categorized as either positive or negative for colorectal cancer in a first-degree relative. Mesalamine use—this was analyzed as a single entity and was also subcategorized into sulfasalazine and nonsulfasalazine medication. Patients were classified according to the cumulative number of years on the drug (never or ⬍3 months in total, between 3 months and 10 years, and ⬎10 years). Azathioprine use—this was similarly categorized into never or ⬍1 year in total, between 1 and 5 years, and ⬎5 years’ use. Folate supplements—the patient was categorized as taking folate supplements where there was evidence of such in the patient’s medical record. Where there was no evidence of this, the patient was categorized as not having taken folate supplements.

All analyses were initially performed on the entire cohort of patients. Because of the continuing debate in the literature about the significance of adenoma-like masses in colitis,4 – 8 we also performed a subanalysis after excluding all cases whose dysplasia had been considered at the time to be adenoma-like. A final subanalysis was performed solely on cancer cases and their respective controls.

February 2004

INFLAMMATION AND COLORECTAL CANCER IN UC

Table 1. Patient Demographics Variable Sex, n (%) Male Female Median age at onset of UC, yr (range) Median duration of UC, yr (range) Patients with pancolitis, n (%)a Total no. colonoscopies (median per patient; range) Median year of index colonoscopy (range)

Cases (n ⫽ 68)

Controls (n ⫽ 136)

39 (57) 29 (43) 33 (6–62)

78 (57) 58 (43) 33 (8–65)

21 (5–52)

23 (7–55)

133 (96.3)

67 (98.5)

407 (5; 1–17)

1990 (1973–2001)

810 (5; 1–13)

1989 (1972–2001)

UC, ulcerative colitis. was defined as histological inflammation at or proximal to the ascending colon. aPancolitis

Statistics Statistical analyses were performed with conditional logistic regression, allowing for grouping of the cases and controls. Univariate analysis was performed on each variable. Factors that showed some evidence of an effect on case-control status from the univariate analyses (P ⬍ 0.2) were then analyzed in a multivariate analysis, and a backwards selection procedure was used to retain the significant risk factors. Nonparametric (Spearman) rank correlation was used to assess correlations between the histological and colonoscopic inflammation scores.

Results Study Population Sixty-eight patients developed colorectal neoplasia while on surveillance between January 1, 1988, and January 1, 2002. The control population therefore comprised 136 matched patients. The median number of surveillance colonoscopies per patient was 5. Overall, data from more than 10,000 biopsy samples (from 9736 colonic segments) at 1217 colonoscopies over 2168 patient-years were studied. There were no significant differences in patient demographics (Table 1). Analysis of the 68 cases showed that the most advanced neoplasia to be documented was colorectal cancer in 14 cases (2 Dukes’ A, 3 Dukes’ B, 7 Dukes’ C, and 2 disseminated malignancies), high-grade dysplasia in 8 patients, and low-grade dysplasia in 14 patients. Thirtytwo patients developed what were believed to be adenomas within inflamed colitic mucosa. Of these, 21 were classified as tubular adenomas (16 with mild, 4 with moderate, and 1 with severe dysplasia), 6 were classified as tubulovillous adenomas (5 with mild and 1 with moderate dysplasia), and 5 were classified as villous ad-

453

enomas (3 with mild and 2 with moderate dysplasia). No cancers were detected at the index colonoscopy, whereas 4 patients had dysplasia detected during the index procedure. The site of the most severe example of neoplasia was the rectum (26.5%), sigmoid (26.5%), descending colon (4.4%), splenic flexure (1.5%), transverse colon (11.8%), hepatic flexure (2.9%), ascending colon (10.3%), and cecum (7.4%). In 6 cases (8.8%), the site was not clearly recorded. Univariate Analysis Univariate analysis showed a highly significant correlation between both the colonoscopic and histological inflammation scores and the risk of colorectal neoplasia (P ⫽ 0.001 and P ⬍ 0.001, respectively). For colonoscopic inflammation, a 1-unit increase in the score increased the odds of colorectal neoplasia by 2.54. A 1-unit increase in the histological inflammation score increased the odds of colorectal neoplasia by a factor of 5. There was a strong positive correlation between these 2 scores (r ⫽ 0.69; 95% confidence interval, 0.60 – 0.75; P ⬍ 0.0001; Spearman rank). Univariate analysis results are shown in Table 2. None of the other factors was found to significantly influence the occurrence of colorectal neoplasia. Patients with PSC showed a trend toward an increased risk of colorectal neoplasia (odds ratio, 4.00; P ⫽ 0.11). Because there was a (nonsignificant) trend for an increased number of years of mesalamine intake and an increased risk of colorectal neoplasia and because there has been previous concern in the medical literature that folate inhibition by sulfasalazine might predispose to cancer,9 we performed a separate analysis by subcategorizing mesalamine drug intake into sulfasalazine and nonsulfasalazine medication (Table 3). By this subanalysis, nonsulfasalazine mesalamine drug use showed a nonsignificant protective trend with increasing duration of use with an odds ratio of 0.79 for up to 10 years’ use and 0.65 for ⬎10 years’ use. Conversely, sulfasalazine medication showed no effect for 3 months to 10 years of use compared with ⬍3 months of sulfasalazine (odds ratio, 0.97), but it showed a deleterious effect on the neoplasia risk for patients who took the drug for ⬎10 years (odds ratio, 1.58; nonsignificant). Azathioprine use had a (nonsignificant) protective effect on cancer risk, with odds ratios of 0.34 for up to 5 years’ use and 0.73 for ⬎5 years’ use. Folate supplement use was also protective (odds ratio, 0.4; nonsignificant). However, this should be interpreted with caution, because the small number of patients with documented evidence of folate supplementation (6 patients) may in part reflect poor documentation of nonprescribed medication.

454

RUTTER ET AL.

GASTROENTEROLOGY Vol. 126, No. 2

Table 2. Results of Univariate Analysis Variable scorea

Colonoscopy inflammation Histological inflammation scorea Family history of CRC (%) No Yes PSC (%) No Yes All mesalamine use (%) Neverb Up to 10 yr ⬎10 yr use Azathioprine use (%) Neverc 1–5 yr use ⬎5 yr use Folate supplement (%) No Yes Smoking status (%) Never Ex Current

Controls

Cases

Odds ratio (95% confidence interval)

P value

1.89 (0.52) 2.05 (0.41)

2.22 (0.78) 2.38 (0.56)

2.54 (1.45–4.44) 5.13 (2.36–11.14)

0.001 ⬍0.001

107 (86) 18 (14)

52 (89) 7 (12)

1 1.09 (0.40–2.94)

0.17

124 (98) 2 (2)

60 (94) 4 (6)

1 4.00 (0.73–21.84)

0.11

13 (10) 37 (27) 85 (63)

3 (4) 16 (24) 49 (72)

1 1.74 (0.45–6.70) 2.38 (0.67–8.54)

0.32

99 (73) 17 (13) 20 (15)

56 (82) 4 (6) 8 (12)

1 0.34 (0.09–1.25) 0.73 (0.30–1.78)

0.22

131 (96) 5 (4)

66 (99) 1 (1)

1 0.40 (0.05–3.42)

0.40

69 (56) 46 (37) 9 (7)

34 (62) 19 (35) 2 (4)

1 0.58 (0.24–1.40) 0.43 (0.08–2.23)

0.37

CRC, colorectal cancer; PSC, primary sclerosing cholangitis. aMean (SD); odds ratio is for a 1-unit increase in score. bNever or ⬍3 months’ cumulative use. cNever or ⬍1 year’s cumulative use.

Analysis of smoking status showed a nonsignificant trend toward a protective effect for current smokers over ex-smokers and for ex-smokers over patients who had never smoked. Because of this trend and because nicotine might have an anti-inflammatory effect on colitic mucosa,10 we recategorized patients into those who had smoked since the onset of their colitis and those who had either stopped smoking before the onset of colitis or had never smoked (Table 4). However, the refined analysis, which theoretically should have been more likely to highlight a beneficial effect of smoking, showed no obvious trend (odds ratio, 0.88; P ⫽ 0.8). Multivariate Analysis Multivariate analysis was performed to examine the joint effect of the explanatory variables on case-

control status. After adjusting for the other explanatory variables, there was a highly significant effect of the histological inflammation score on the occurrence of case-control status (P ⬍ 0.001). A 1-unit increase in the score increased the odds of colorectal neoplasia by a factor of 4.69 (95% confidence interval, 2.10 –10.48; P ⬍ 0.001). No other factor reached statistical significance by multivariate analysis. Subanalysis Excluding Adenoma Cases Univariate and multivariate analyses were repeated for cases with nonadenomatous dysplasia or cancer and their matched controls only (n ⫽ 108). Univariate results are summarized in Table 5. Again, on univariate analysis, higher colonoscopy and histological inflammation scores were the only factors

Table 3. Mesalamine Subcategorization Variable Sulfasalazine use (%) Nevera Up to 10 yr ⬎10 yr use Nonsulfasalazine mesalamine use (%) Nevera Up to 10 yr ⬎10 yr use aNever

or ⬍3 months’ cumulative use.

Controls

Cases

Odds ratio (95% confidence interval)

P value

33 (25) 43 (32) 58 (43)

14 (21) 17 (25) 37 (54)

1 0.97 (0.41–2.26) 1.58 (0.71–3.51)

0.30

71 (53) 41 (31) 21 (16)

41 (60) 19 (28) 8 (12)

1 0.79 (0.40–1.55) 0.65 (0.26–1.62)

0.58

February 2004

INFLAMMATION AND COLORECTAL CANCER IN UC

Table 4. Refined Smoking History Variable

Controls

Cases

Odds ratio (95% confidence interval)

Smoking since UC onset (%) No Yes

98 (86) 14 (14)

45 (88) 6 (12)

1 0.88 (0.32–4.42)

P value

0.80

UC, ulcerative colitis.

found to significantly increase the occurrence of neoplasia (P ⫽ 0.001 and P ⬍ 0.001, respectively). For colonoscopic inflammation, a 1-unit increase in the score increased the odds of colorectal neoplasia by ⬎5. A 1-unit increase in the histological inflammation score increased the odds of colorectal neoplasia by a factor of 7. After multivariate analysis, only the histological inflammation

455

score remained significant, with an odds ratio of 5.40 (95% confidence interval, 1.82–16.02; P ⫽ 0.002). Subanalysis of Cancer Cases Only Univariate and multivariate analyses were repeated for cases with cancer and their matched controls only (n ⫽ 42). Univariate results are summarized in Table 6. Because all patients had taken a mesalamine drug, it was decided to combine the “never taken mesalamine/less than 3 months’ use” group with the “3month to 10-year use” group. Similarly, there were no patients who had taken azathioprine for 1 to 5 years, so these subjects were combined with the “more than 5 years” group. On univariate analysis, only the histological inflammation score was found to significantly influence the case-control status (P ⫽ 0.03). A 1-unit increase in the

Table 5. Univariate Analysis for Nonadenomatous Dysplasia and Cancer Cases With Controls Only Variable scorea

Colonoscopy inflammation Histological inflammation scorea PSC (%) No Yes Family history of CRC (%) No Yes Smoking since UC onset (%) No Yes Smoking status (%) Never Ex Current Sulfasalazine use (%) Neverc Up to 10 yr ⬎10 yr use Nonsulfasalazine mesalamine use (%) Neverc Up to 10 yr ⬎10 yr use All mesalamine use (%) Nevera Up to 10 yr ⬎10 yr use Azathioprine use (%) Neverd 1–5 yr use ⬎5 yr use Folate supplement (%) No Yes

Controls

Cases

Odds ratio (95% confidence interval)

P value

1.96 (0.51) 2.10 (0.43)

2.54 (0.84) 2.59 (0.59)

5.33 (2.00–14.19) 6.98 (2.42–20.16)

0.001 ⬍0.001

66 (97) 2 (3)

31 (89) 4 (11)

1 4.00 (0.73–21.83)

0.11

61 (91) 6 (9)

30 (94) 2 (6)

1 1.33 (0.22–7.98)

0.75

52 (83) 11 (17)

31 (100) 0 (0)

—b

39 (57) 23 (33) 7 (10)

24 (73) 9 (27) 0 (0)

—b

19 (26) 18 (25) 35 (49)

7 (19) 11 (31) 18 (50)

1 1.59 (0.53–4.79) 1.36 (0.48–3.81)

0.70

36 (51) 24 (34) 11 (15)

23 (64) 8 (22) 5 (14)

1 0.52 (0.20–1.34) 0.63 (0.18–2.21)

0.38

8 (11) 15 (21) 49 (68)

1 (3) 9 (25) 26 (72)

1 4.28 (0.48–37.97) 3.89 (0.48–31.62)

0.42

53 (74) 9 (13) 10 (14)

32 (89) 3 (8) 1 (3)

1 0.47 (0.09–2.43) 0.19 (0.02–1.48)

0.20

69 (96) 3 (4)

35 (97) 1 (3)

1 0.67 (0.07–6.41)

0.73

CRC, colorectal cancer; UC, ulcerative colitis. (standard deviation); odds ratio is for 1 unit increase in score. bOdds ratios could not be calculated because there were no subjects in some categories. cNever or ⬍3 months’ cumulative use. dNever or ⬍1 year’s cumulative use. aMean

456

RUTTER ET AL.

GASTROENTEROLOGY Vol. 126, No. 2

Table 6. Univariate Analysis for Cancer Cases With Controls Only Variable scorea

Colonoscopy inflammation Histological inflammation scorea PSC (%) No Yes Family history of CRC (%) No Yes Smoking since UC onset (%) No Yes Smoking status (%) Never Ex Current Sulfasalazine use (%) Neverc Up to 10 yr ⬎10 yr use Nonsulfasalazine mesalamine use (%) Neverc Up to 10 yr ⬎10 yr use All mesalamine use (%) Never/ⱕ10 yr ⬎10 yr use Azathioprine use (%) Neverd ⱖ1 yr use Folate supplement (%) No Yes

Controls

Cases

Odds ratio (95% confidence interval)

P value

2.00 (0.55) 2.09 (0.44)

2.45 (0.98) 2.54 (0.60)

2.62 (0.84–8.21) 6.33 (1.24–32.33)

0.10 0.03

27 (96) 1 (4)

12 (86) 2 (14)

1 4.00 (0.36–44.11)

0.26

24 (92) 2 (7)

12 (92) 1 (8)

1 1.22 (0.09–11.03)

1.00

22 (81) 5 (19)

12 (100) 0 (0)

—b

19 (70) 4 (15) 4 (15)

10 (77) 3 (23) 0 (0)

—b

9 (32) 8 (29) 11 (39)

1 (7) 5 (36) 8 (57)

1 4.89 (0.47–50.96) 6.59 (0.64–67.92)

0.27

16 (57) 8 (29) 4 (14)

11 (79) 2 (14) 1 (7)

1 0.32 (0.05–2.02) 0.32 (0.03–3.32)

0.40

11 (39) 17 (61)

3 (21) 11 (79)

1 2.10 (0.52–8.43)

0.29

20 (71) 8 (29)

13 (93) 1 (7)

1 0.22 (0.03–1.87)

0.17

27 (96) 1 (4)

13 (93) 1 (7)

1 2.00 (0.13–31.98)

0.62

CRC, colorectal cancer; UC, ulcerative colitis. (SD); odds ratio is for a 1-unit increase in score. bOdds ratios could not be calculated because there were no subjects in some categories. cNever or ⬍3 months’ cumulative use. dNever or ⬍1 year’s cumulative use. aMean

score increased the odds of colorectal cancer 6-fold. Multivariate analysis indicated that, again, only the histological inflammation score influenced the occurrence of colorectal cancer. Because this was the only variable in the final model, the effect size and significance level were the same as for the univariate analysis (odds ratio, 6.33; P ⫽ 0.03).

Discussion Severity of Inflammation It is generally assumed that the excess colorectal cancer risk in ulcerative colitis is predominantly acquired,2,11 although genetic factors such as a family history of colonic cancer further increase this risk.12 The alternative hypothesis of a genetic link between colorectal cancer and ulcerative colitis1 seems unlikely in light of a recent study showing that first-degree relatives of

those with ulcerative colitis are not at increased risk of colorectal cancer.3 It is biologically plausible that the excess cancer risk is secondary to chronic inflammation, and it is recognized that duration of colitis is an important risk factor for colorectal cancer. However, until now is has not been possible to show an association between the severity of colonic inflammation and cancer risk in ulcerative colitis. The quantification of disease severity has remained a major problem facing any such study. Edwards and Truelove13 reported increased rates of cancer in patients with chronic continuous symptoms over those with chronic intermittent symptoms, but 2 subsequent case-control studies that used the frequency of hospital visits, frequency of colonic imaging, and frequency of clinical exacerbations to categorize disease severity failed to show an association.14,15 Indeed, even more complex clinical indices of disease activity, such as the Powell–Tuck Activity Score, correlate only modestly

February 2004

with colonoscopic and histological inflammation16 and are therefore relatively poor surrogate markers for colonic inflammation. Colonic inflammation varies considerably, both over time and throughout the length of the colon. Quantifying the overall inflammation for an individual is difficult. In our study, inflammation grading was derived from a composite score from segmental biopsy samples taken throughout the colon during surveillance. In total, we assessed more than 10,000 biopsy specimens from 1217 non–symptom-driven surveillance colonoscopies. We believe that this approach has given us the most representative record possible of each patient’s colonic inflammation over time and throughout the colorectum. Using this method, we have been able to show for the first time that increasing severity of colonic inflammation is associated with an increase in the rate of colorectal neoplasia in ulcerative colitis. Biases and Drawbacks of the Study Because of the considerable timescales involved in the development of colorectal cancer in ulcerative colitis, it would have been impractical to perform a prospective trial. Thus, we devised a retrospective case-control study. Hence, we were reliant on accurate data documentation at the time of each colonoscopic procedure. Patients with ulcerative colitis can have different severity of inflammation over time. For the purpose of this study, we assumed that the inflammation seen at that patient’s programmed surveillance colonoscopy was representative of the inflammation in the intervening period between examinations. Although this might not necessarily be the case, we believed that there was no better means to assess the chronology of macroscopic and histological inflammation severity. Also, by scoring more than 1200 colonoscopies, we believed that the chance of bias was minimal. To devise the histological and colonoscopic scoring system, we consulted with our hospital’s histopathologists and endoscopists to obtain simple macroscopic and histological severity of inflammation scales that would conform with the data available on the prospectively generated colonoscopy and histology reports. As such, the scores were not validated. Our study did not specifically look at the effect of disease extent on neoplasia risk, because all patients entered into the St. Mark’s Hospital ulcerative colitis surveillance program had, by definition, macroscopically extensive disease proximal to the splenic flexure. Because previous studies have clearly determined extent to be a risk factor,17 we believed that it was important to match for disease extent to avoid potential bias. An interesting observation, however, was that 200 of the 204 patients in

INFLAMMATION AND COLORECTAL CANCER IN UC

457

the study had had microscopic inflammation at or proximal to the ascending colon at some time during surveillance. Thus, it would seem that virtually all patients with macroscopic inflammation proximal to the splenic flexure have had microscopic inflammation involving the entire colon at some point in time. Drugs and Inflammation One possible mechanism by which inflammation might promote carcinogenesis in colitis is through altered prostaglandin metabolism.11 Inflammation induces increased expression of cyclooxygenase 2, which results in depleted proapoptotic intracellular unesterified arachidonic acid and increased expression of the antiapoptotic Bcl-2.18 These effects, plus the fact that cyclooxygenase 2 is also associated with angiogenesis,19 create a microenvironment favorable for cancer development. Other possible explanations may emerge in due course. Given this possible mechanism for inflammationinduced cancer, one might expect patients taking antiinflammatory mesalamine drugs to show reduced rates of colorectal neoplasia. However, our results show a (nonsignificant) trend toward an increased risk of colorectal neoplasia in those taking mesalamine drugs. Another recent study of 143 patients over 20 years similarly failed to show a protective effect for sulfasalazine,20 although 3 other retrospective studies14,15,21 (2 from the same center) concluded that mesalamine reduced colorectal cancer risk by up to 75%. One possible reason for these conflicting conclusions is that for patients taking sulfasalazine, which is a competitive inhibitor of folate absorption, the anti-inflammatory effect is outweighed by the effect of folate deficiency, which has been associated with dysplasia development.9 Our results support this dichotomy: on subanalysis by mesalamine drug type, patients taking sulfasalazine showed a trend toward increased neoplasia risk with increasing years of use, whereas those taking nonsulfasalazine mesalamine drugs showed a trend toward reduced neoplasia rates with increasing years of use. This finding has important implications for the physician selecting the most appropriate long-term medication for their patients and indicates that in terms of neoplasia prevention, it is preferable for patients taking long-term mesalamine drugs to take nonsulfasalazine compounds; if there is a specific reason why sulfasalazine is required, they should consider folate supplements also. A recent study in a primary care setting reported that 38% of patients with extensive colitis were not taking aminosalicylate drugs,22 suggesting that there is significant room for improvement in terms of cancer chemoprevention.

458

RUTTER ET AL.

The finding that both nonsulfasalazine mesalamine drugs and azathioprine showed a trend toward a reduced risk of colorectal neoplasia supports the theory that it is the end-point severity of colonic inflammation that is important in terms of neoplasia risk (and, thus, that neoplasia is a corollary of inflammation), as opposed to the patient’s underlying genotype. Because colonic inflammation was the only significant factor on multivariate analysis, it is logical for the physician to aim to minimize colonic inflammation by whichever drugs are necessary, in the hope (supported by our evidence) that by reducing the severity of colonic inflammation, the patient’s risk of colorectal cancer is reduced. Primary Sclerosing Cholangitis This study supports other reports showing that patients with PSC are at increased risk of colorectal cancer23–25: patients with PSC in our study had an odds ratio of 4 for colorectal neoplasia. However, because relatively few patients in our cohort had PSC, these results did not achieve statistical significance. Our findings are also interesting with regard to the etiology of the increased risk in PSC. One widely held hypothesis for the increased cancer risk is that patients with PSC often have mild inflammation and might therefore have had mild subclinical disease for many years before their diagnosis of colitis, thus having had a longer disease duration than is clinically apparent and, hence, a higher cancer risk. Our study’s finding would suggest that the mild colonic inflammation that accompanies PSC might not be sufficient to explain the excess cancer risk in these patients and that it might therefore be more likely that PSC is an independent risk factor for cancer in colitis. Smoking There is now increasing evidence that cigarette smoking promotes colorectal adenoma and cancer development in the noncolitic population, although there is a considerable lag time of several decades.26,27 In our study, a history of smoking did not increase the risk of colorectal neoplasia in ulcerative colitis. In fact, the initial analysis suggested a trend toward a protective effect for current smokers over ex-smokers and for exsmokers over those who have never smoked. A previous retrospective case-control study found a similar trend.14 One possible explanation for a protective effect is the anti-inflammatory action of cigarette smoking on the colon. To examine this more closely, we reanalyzed data after categorizing patients into those who had smoked since the onset of their colitis (and thus had the potential to benefit from an anti-inflammatory effect of cigarette smoking on their colitis) and those who had not. By this

GASTROENTEROLOGY Vol. 126, No. 2

analysis, the trend toward a protective effect for cigarette smoking was reduced (odds ratio, 0.88; not significant); thus, we were unable to find evidence to support this hypothesis. It also remains possible that cigarette smoking is detrimental but that our study was too short to allow for a similar induction time of several decades for sporadic colorectal cancer. Colectomy Rates Our finding that increasing severity of inflammation increases the risk of colorectal neoplasia helps to explain the considerable disparity in the reported cumulative incidences of colorectal cancer between centers. In centers such as Denmark, colectomy rates for chronic colitic symptoms are high, with figures of ⬎30% by 25 years’ colitis duration in 1 study.28 These centers thus remove the colons and, hence, the cancer risk from the patients with more severe colonic inflammation, who would have an above-average risk of developing colorectal cancer. Our findings might also help to explain anecdotal reports of more cancers occurring in specialist centers as opposed to district general hospitals: tertiary referral centers are more likely to treat patients with more severe disease and, thus, a higher than average colorectal cancer risk. This has important implications for the cost-effectiveness of cancer surveillance programs in ulcerative colitis in different health-care settings. Crohn’s Inflammation Although we did not study patients with Crohn’s colitis in our study, our finding that inflammation is important for neoplasia risk is also consistent with increasing evidence that the colorectal cancer risk in Crohn’s colitis is similar to that in ulcerative colitis when the duration of disease is similar.11 Moreover, studies show that the increased colorectal cancer risk is evident only in patients with Crohn’s colitis, as opposed to small-bowel disease.29

Conclusion We have shown that in patients with long-standing, extensive ulcerative colitis, the severity of colonic inflammation is an important determinant of the risk of colorectal neoplasia. In the future, it might be possible to adjust a patient’s surveillance frequency according to the severity of ongoing active inflammation.

References 1. Rhodes JM. Unifying hypothesis for inflammatory bowel disease and associated colon cancer: sticking the pieces together with sugar. Lancet 1996;347:40 – 44.

February 2004

2. Shanahan F. Relation between colitis and colon cancer. Lancet 2001;357:246 –247. 3. Askling J, Dickman PW, Karlen P, Brostrom O, Lapidus A, Lofberg R, Ekbom A. Colorectal cancer rates among first-degree relatives of patients with inflammatory bowel disease: a population-based cohort study. Lancet 2001;357:262–266. 4. Connell WR, Lennard-Jones JE, Williams CB, Talbot IC, Price AB, Wilkinson KH. Factors affecting the outcome of endoscopic surveillance for cancer in ulcerative colitis. Gastroenterology 1994; 107:934 –944. 5. Torres C, Antonioli D, Odze RD. Polypoid dysplasia and adenomas in inflammatory bowel disease: a clinical, pathologic, and follow-up study of 89 polyps from 59 patients. Am J Surg Pathol 1998;22:275–284. 6. Engelsgjerd M, Farraye FA, Odze RD. Polypectomy may be adequate treatment for adenoma-like dysplastic lesions in chronic ulcerative colitis. Gastroenterology 1999;117:1288 –1294; discussion 1488 –1491. 7. Rubin PH, Friedman S, Harpaz N, Goldstein E, Weiser J, Schiller J, Waye JD, Present DH. Colonoscopic polypectomy in chronic colitis: conservative management after endoscopic resection of dysplastic polyps. Gastroenterology 1999;117:1295–1300. 8. Bernstein CN. ALMs versus DALMs in ulcerative colitis: polypectomy or colectomy? Gastroenterology 1999;117:1488 –1492. 9. 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;97:255–259. 10. Motley RJ, Rhodes J, Williams G, Tavares IA, Bennett A. Smoking, eicosanoids and ulcerative colitis. J Pharm Pharmacol 1990;42: 288 –289. 11. Rhodes JM, Campbell BJ. Inflammation and colorectal cancer: IBD-associated and sporadic cancer compared. Trends Mol Med 2002;8:10 –16. 12. 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. 13. Edwards FC, Truelove SC. The course and prognosis of ulcerative colitis: Part IV: carcinoma of the colon. Gut 1964;5:15–22. 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;107:117–120. 15. Eaden J, Abrams K, Ekbom A, Jackson E, Mayberry J. Colorectal cancer prevention in ulcerative colitis: a case-control study. Aliment Pharmacol Ther 2000;14:145–153. 16. Gomes P, du Boulay C, Smith CL, Holdstock G. Relationship between disease activity indices and colonoscopic findings in patients with colonic inflammatory bowel disease. Gut 1986;27: 92–95. 17. Ekbom A, Helmick C, Zack M, Adami HO. Ulcerative colitis and

INFLAMMATION AND COLORECTAL CANCER IN UC

18.

19. 20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

459

colorectal cancer: A population-based study. N Engl J Med 1990; 323:1228 –1233. Tsujii M, DuBois RN. Alterations in cellular adhesion and apoptosis in epithelial cells overexpressing prostaglandin endoperoxide synthase 2. Cell 1995;83:493–501. Gately S. The contributions of cyclooxygenase-2 to tumor angiogenesis. Cancer Metastasis Rev 2000;19:19 –27. Lindberg BU, Broome U, Persson B. Proximal colorectal dysplasia or cancer in ulcerative colitis: The impact of primary sclerosing cholangitis and sulfasalazine: results from a 20-year surveillance study. Dis Colon Rectum 2001;44:77– 85. Moody GA, Jayanthi V, Probert CS, Mac K, 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. Rubin G, Hungin AP, Chinn D, Dwarakanath AD, Green L, Bates J. Long-term aminosalicylate therapy is under-used in patients with ulcerative colitis: a cross-sectional survey. Aliment Pharmacol Ther 2002;16:1889 –1893. Jayaram H, Satsangi J, Chapman RW. Increased colorectal neoplasia in chronic ulcerative colitis complicated by primary sclerosing cholangitis: fact or fiction? Gut 2001;48:430 – 434. Broome U, Lindberg G, Lofberg R. Primary sclerosing cholangitis in ulcerative colitis—a risk factor for the development of dysplasia and DNA aneuploidy? Gastroenterology 1992;102: 1877–1880. Brentnall TA, Haggitt RC, Rabinovitch PS, Kimmey MB, Bronner MP, Levine DS, Kowdley KV, Stevens AC, Crispin DA, Emond M, Rubin CE. Risk and natural history of colonic neoplasia in patients with primary sclerosing cholangitis and ulcerative colitis. Gastroenterology 1996;110:331–338. Giovannucci E, Colditz GA, Stampfer MJ, Hunter D, Rosner BA, Willett WC, Speizer FE. A prospective study of cigarette smoking and risk of colorectal adenoma and colorectal cancer in U.S women. J Natl Cancer Inst 1994;86:192–199. Giovannucci E, Rimm EB, Stampfer MJ, Colditz GA, Ascherio A, Kearney J, Willett WC. A prospective study of cigarette smoking and risk of colorectal adenoma and colorectal cancer in U.S men. J Natl Cancer Inst 1994;86:183–191. Langholz E, Munkholm P, Davidsen M, Binder V. Colorectal cancer risk and mortality in patients with ulcerative colitis. Gastroenterology 1992;103:1444 –1451. Ekbom A, Helmick C, Zack M, Adami HO. Increased risk of large-bowel cancer in Crohn’s disease with colonic involvement. Lancet 1990;336:357–359.

Received July 17, 2003. Accepted October 23, 2003. Address requests for reprints to: Matthew Rutter, MBBS, MRCP (UK), Department of Gastroenterology, University Hospital of North Tees, Stockton-on-Tees, Teesside TS19 8PE, England. e-mail: Matt.rutter@ nth.nhs.uk; fax: (44) 0-1642-383289.