Risk factors for proximal disease extension and colectomy in left-sided ulcerative colitis

Abstract

Objective

The primary objective of this study was to assess proximal disease extension of ulcerative colitis (UC) over time, with disease behaviour pattern and risk factors for proximal disease extension and colectomy as secondary aims.

Methods

All cumulative incident cases diagnosed with UC at the Academic Medical Center between January 1990 and December 2009 were studied. The cumulative risk of colectomy was calculated by Kaplan-Meier analysis. The Cox proportional hazards regression was used to identify risk factors associated with proximal disease extension and colectomy.

Results

In total, 506 UC patients were included with a median age of 33 years (IQR 23–41) at diagnosis. Ninety-five (18.8%) patients underwent colectomy during follow-up. Median follow-up was 10 years (IQR 5–15). Initial disease extent was evaluable in 416 patients, of whom 142 (34.1%) had proctitis, 155 (37.3%) left-sided colitis and 119 (28.6%) pancolitis. Proximal disease extension was observed in 120 (28.8%) patients during follow-up (51 proctitis to left-sided colitis, 39 proctitis to extensive colitis and 30 left-sided to extensive colitis). Disease behaviour was evaluable in 378 patients, of whom 244 (64.6%) had less than one relapse per year. Younger age at diagnosis (HR 0.98, 95% CI 0.96–0.99) and continuous active disease (HR 2.18, 95% CI 1.27–3.73) were independent risk factors for proximal disease extension. The cumulative risk of colectomy did not change over time between patients diagnosed before and after the year 2000 (p = 0.341). Continuous active disease (HR 7.05, 95% CI 4.23–11.77), systemic steroids (HR 3.25, 95% CI 1.37–7.71) and cyclosporine treatment (HR 2.80, 95% CI 1.66–4.72) were independent risk factors for colectomy, whereas proctitis at diagnosis (HR 0.43, 95% CI 0.22–0.86) carried a lower risk.

Conclusion

In one-third of UC patients, left-sided disease at diagnosis will extend proximally during 10 years of follow-up. Proximal disease extension was not a risk factor for colectomy, but the risk of colectomy is rather determined by continuous disease activity, and use of systemic steroids and cyclosporine.

Introduction

Ulcerative colitis (UC) is diagnosed in approximately 9–20 patients per 100,000 inhabitants per year in Northern Europe and North America and poses a formidable burden on the lives of young adults.¹ The clinical presentation at onset of the disease and the subsequent disease course vary significantly among patients. In general, the disease course of UC is characterised by a relapsing and remitting behaviour, but up to 6% of patients experience chronic continuous symptoms.² Furthermore, distal colitis may progress towards the proximal colon, whereas extensive UC may regress over time. This pattern is difficult to predict and may influence the prognosis and the necessity for colectomy.

The European Crohn’s and Colitis Organisation (ECCO) guidelines recommend the use of the Montreal Classification for defining the distribution of disease to describe the maximal proximal disease extent of inflammation seen at colonoscopy.³ Using this classification system, earlier series have shown that approximately 30%–50% of UC patient have disease confined to the rectum, 20%–30% have left-sided colitis and 20%–30% have disease that extends beyond the hepatic flexure (then named ‘extensive colitis’ or ‘pancolitis’).^4,5 More recent studies have shown a preponderance for left-sided colitis and suggested that an initial presentation with extensive colitis could be a risk factor for colectomy.^6,7 Knowledge about the extent of the disease is essential to determine the optimal treatment strategy (e.g. appropriateness of topical treatment) and estimate the need for colectomy in the near future.

Over the last two decades, medical treatment has evolved from suppression of symptoms, mainly with sulfasalazine, 5-aminosalycates and glucocorticosteroids, to more immunomodulatory and targeted therapies such as anti-tumour necrosis factor (anti-TNF) and integrin antibodies.⁸ Since most of the previously mentioned studies have been performed before the era of biologicals, it remains unknown to what extent disease course and colectomy rates have changed in more recent years. Furthermore, studies that have investigated disease behaviour and extent over time are limited in number. It has been suggested that patients who experience chronic active symptoms or have disease that extends proximally over time may have an aggressive phenotype and an increased risk of colectomy. If the behaviour of UC and the risk for colectomy could be predicted, patients could be treated more appropriately earlier during their disease course.

Therefore, the objectives of this study were to assess the progression of extent of disease and disease behaviour patterns and to identify prognostic risk factors for proximal disease extension and colectomy.

Methods

Patient population

In this retrospective study, we included all UC patients treated at the Academic Medical Center (AMC) in Amsterdam, the Netherlands, between January 1990 and December 2009. Patients in the AMC are registered at initial presentation by diagnosis code. Patients diagnosed with unclassified inflammatory bowel disease (IBD-U), Crohn’s disease (CD), microscopic colitis or infectious proctitis were excluded. Patients in whom the date of diagnosis was unknown were also excluded. Medical records were reviewed to obtain additional data. The demographic variables that were extracted included gender, age, smoking habits (current, stopped, never), body mass index (BMI), comorbidity grouped by organ system and extra-intestinal manifestations. Disease-specific data on disease localisation, behaviour, and medical and surgical treatment were also registered.

Disease characteristics

All colonoscopy reports were reviewed to determine the UC disease extent and activity at diagnosis and during follow-up. Disease extent was categorised according to the Montreal Classification: E1 (proctitis): distal to the sigmoid, E2 (left-sided): distal to the splenic flexure, and E3 (extensive): proximal to the splenic flexure.⁹ A colonoscopy was defined as inconclusive if the upper limit of inflammation was not reached. An exception was made when the scope reached the hepatic flexure and the upper limit was not seen; these cases were scored as extensive colitis.

Proximal disease extension was defined as inflammation extending proximally from proctitis to left-sided colitis, from proctitis to extensive colitis or from left-sided colitis to extensive colitis. A relapse was defined as the occurrence of blood and mucus in the stool. Disease behaviour has been previously classified in four different patterns in the IBSEN study, in which the patients were asked to choose the curve that best described the course of their disease during this period.¹⁰ In the current study, we determined the disease behaviour based on patient charts and categorised it according to the periods of exacerbation and remission as shown in Figure 1: (a) remission after diagnosis, (b) less than one relapse per year, (c) more than one relapse per year and (d) continuous activity.

Figure 1.

Disease behavior in ulcerative colitis (UC) patients: (a) remission after diagnosis, (b) less than one relapse per year, (c) more than one relapse per year and (d) continuous activity.

Medical treatment

We registered the use of 5-aminosalycilic acid (5-ASA), glucocorticosteroids (ever received topical and systemic treatment), immunomodulators (azathioprine, 6-Tioguanine, 6-Mercaptopurine and methotrexate), cyclosporine and various biologicals.

Outcome measures

Our primary outcome measure was the incidence of proximal disease extension. The secondary outcome measures included disease behaviour, risk factors for proximal disease extension and risk factors for colectomy.

Statistical analysis

Data were presented as counts and percentages or medians and interquartile ranges (IQR). For the purpose of the Kaplan-Meier and the Cox proportional hazard analyses, a ‘time to extension’ and ‘time to colectomy’ variable was created. Patients were censored if at the end of the follow-up period no proximal extension or colectomy had occurred, if they had died or were lost to follow-up. Kaplan-Meier curves of patients who had been diagnosed with UC before and after the year 2000 were compared by log rank test to evaluate whether the rates have changed over time. The year 2000 was chosen since this was half way between the time span of the current cohort, which coincided with the advent of the biological treatment era. The Cox proportional hazards regression was used to identify risk factors associated with proximal disease extension. Hereby, patients with extensive disease at diagnosis were excluded from this analysis. All medicinal variables were scored as ‘yes’ when patients were treated before proximal disease extension had occurred. The Cox proportional hazards regression was also used to identify risk factors associated with colectomy. In this analysis, all patients who were indicated for colectomy due to dysplasia or cancer were excluded from the analysis. In univariable analysis, variables with a p value of less than 0.2 were considered for multivariable analysis. Independent risk factors were identified by forward selection and considered to be significant with a two-sided p value of <0.05. Analysis was performed using IBM SPSS for Windows version 22 (SPSS Inc, Chicago, IL, USA).

Results

Study population

In total, 888 patients were identified through the hospital registry and 506 UC patients (49.8% male) were included in the study with a median age of 33 years (IQR 23–41) at diagnosis. Fourteen patients with IBD-U, 80 CD patients, four patients with microscopic colitis, one with infectious proctitis and 283 patients in whom the date of diagnosis was unknown were excluded (Figure 2). The median duration of follow-up was 10 years (IQR 5–15). Eight (1.6%) patients were diagnosed with gastrointestinal cancer during follow-up, of which three were localised in the sigmoid. Eight (1.6%) patients died during follow-up, all due to other causes than UC. The demographic characteristics of the study cohort are presented in Table 1.

Figure 2.

Flow diagram of patients included in the study.

Table 1.

Patient characteristics.

	Total cohort	Colectomy	No colectomy
Total n (%)	506 (100)	95 (18.8)	411 (81.2)
Male	252 (49.8)	51 (53.7)	201 (48.9)
Age at diagnosisa	33 (23–41)	29 (21–39)	32 (23–43)
Age at colectomya	36 (29–43)	–	–
Smoking
Never	248 (58.8)	46 (64.8)	202 (57.5)
Stopped	82 (19.4)	10 (14.1)	72 (20.5)
Current	92 (21.8)	15 (21.1)	77 (21.9)
EIM	69 (15.1)	11 (14.3)	58 (15.2)
PSC	17 (3.7)	6 (7.7)	11 (2.9)
Diabetes	10 (2.0)	0 (0.0)	10 (2.6)
Medication
Systemic steroids	294 (62.2)	71 (86.6)	223 (57.0)
Topical steroids	237 (51.5)	42 (62.7)	195 (49.6)
5-ASA	473 (96.9)	74 (92.5)	399 (97.8)
Immunomodulators	265 (54.6)	63 (75.9)	202 (50.2)
Cyclosporine	60 (12.5)	27 (33.8)	33 (8.3)
Biologicals	104 (21.6)	33 (41.3)	71 (17.7)
Dysplasia	4 (0.8)	4 (4.2)	0 (0.0)
Cancer	8 (1.6)	6 (6.1)	2 (0.5)
Mortality	8 (1.6)	2 (2.1)	6 (1.5)
Follow-up (years)a	10 (5–15)	4 2–9b	11 (7–16)

^aMedian and IQR.

^bFollow-up stopped at the time of colectomy.

Missing values: smoking (n = 84); EIM (n = 48); PSC (n = 47); appendectomy (n = 65); topical steroids (n = 46); systemic steroids (n = 33); 5-ASA (n = 18); Immunomodulators (n = 21); cyclosporine (n = 26); Anti-TNF (n = 24).

EIM: extra-intestinal manifestation; PSC: primary sclerosing cholangitis; 5-ASA: 5-aminosalicylic acid; Anti-TNF: anti-tumour necrosis factor; IQR: interquartile range.

Proximal extension

Extent of disease could not be defined in 90 of 506 (17.7%) patients because colonoscopy reports were missing or contained insufficient details. This resulted in 416 evaluable patients for disease progression. At diagnosis, proctitis was present in 142 of these patients (34.1%). During follow-up, 51 (35.9%) progressed to left-sided colitis and 39 (27.5%) to extensive colitis. Initial left-sided UC was diagnosed in 155 (37.3%) patients. Thirty of these patients (19.4%) progressed to extensive colitis. Extensive disease at primary diagnosis was found in 119 (28.6%) patients (Figure 3).

Figure 3.

Extent of disease at diagnosis and during follow-up (n = 416 evaluable patients with colonoscopy data during follow-up).

Disease behaviour

Disease pattern could be evaluated in 378 out of the 506 patients because of missing data. During follow-up, 244 (64.6%) patients showed a disease pattern as shown in Figure 1(b) with less than one relapse per year, irrespective of extent of disease at primary diagnosis. Continuously active disease was observed in 36 (9.5%) patients (Figure 4).

Figure 4.

Disease behaviour according to disease extent at diagnosis (n = 378 evaluable patients).

Risk factors for proximal disease extension

Univariable analysis identified age at diagnosis, stopped smoking, continuous active disease, systemic steroid treatment, topical steroid treatment, immunomodulators, and biologicals for multivariable regression analysis, based on a p value of <0.2. Multivariable analysis identified younger age at diagnosis (hazard ratio (HR) 0.98, 95% confidence interval (CI) 0.96–0.99) and continuous active disease (HR 2.18, 95% CI 1.27–3.73) as independent risk factors for disease extension (Table 2).

Table 2.

Univariable and multivariable cox regression analysis for the risk of proximal disease extension in UC.

	Univariable cox analysis HR (95%CI)	p value	Multivariable cox analysis HR (95%CI)	p value
Male sex	0.94 (0.66–1.35)	0.742
Age at diagnosis	0.97 (0.96–0.98)	<0.001	0.98 (0.96–0.99)	0.004
Current smoking	0.79 (0.46–1.35)	0.390
Stopped smoking	0.74 (0.48–1.15)	0.182	(−)
EIM	1.06 (0.60–1.85)	0.843
PSC	1.26 (0.40–3.99)	0.690
Diabetes	0.94 (0.30–2.95)	0.912
Continuous active disease	2.36 (1.38–4.02)	0.002	2.18 (1.27–3.73)	0.005
Systemic steroids	1.01 (1.01–1.04)	0.001	(−)
Topical steroids	1.40 (0.95–2.06)	0.090	(−)
5-ASA treatment	1.12 (0.35–3.52)	0.852
Immunomodulators	2.19 (1.49–3.20)	<0.001	(−)
Cyclosporine	1.00 (0.89–1.01)	0.483	(−)
Biologicals	1.65 (1.10–2.53)	0.021	(−)

(−) Indicates a not significant variable that was excluded from multivariable analysis after forward stepwise regression.

UC: ulcerative colitis; EIM: extra-intestinal manifestation, PSC: primary sclerosing cholangitis; 5-ASA: 5-aminosalicylic acid; HR: hazard ratio; CI: confidence interval.

Risk factors for colectomy

In total, 95 (18.8%) of 506 patients underwent colectomy during follow-up with a median time to colectomy of 54 months (IQR 31–111). After five, 10 and 15 years, the cumulative risk of colectomy was 11.7% (95% CI, 8.8–14.6), 17.2% (95% CI, 13.5–20.7) and 20.8% (95% CI, 16.8–25.2), respectively. The cumulative risk of colectomy did not change over time between patients diagnosed before and after the year 2000 (p = 0.34) (Figure 5). Univariable analysis identified the following variables for multivariable analysis (p < 0.2): age at diagnosis, stopped smoking, proctitis at diagnosis, extensive colitis at diagnosis, proximal extension over time, continuous disease activity, systemic steroids, topical steroids, 5-ASA treatment, immunomodulators, cyclosporine and biologicals. Multivariable analysis identified continuously active disease (HR 7.05, 95% CI 4.23–11.77), systemic steroids (HR 3.25, 95% CI 1.37–7.71) and cyclosporine treatment (HR 2.80, 95% CI 1.66–4.72) as independent risk factors for colectomy, whereas proctitis at diagnosis (HR 0.43, 95% CI 0.22–0.86) carried a lower risk of colectomy (Table 3).

Figure 5.

Kaplan-Meier curve for cumulative risk of colectomy before and after the year 2000.

Table 3.

Univariable and multivariable cox regression analysis for the risk of colectomy in UC.

	Univariable cox analysis HR (95%CI)	p value	Multivariable cox analysis HR (95%CI)	p value
Male sex	1.04 (0.68–1.59)	0.867
Age at diagnosis	0.98 (0.97–1.00)	0.069	(−)
Current smoking	0.74 (0.36–1.52)	0.409
Stopped smoking	0.55 (0.27–1.10)	0.091	(−)
EIM	0.77 (0.37–1.62)	0.495
PSC	1.44 (0.45–4.57)	0.541
Diabetes	0.05 (0.01–55.2)	0.399
Proctitisa	0.44 (0.23–0.84)	0.014	0.43 (0.22–0.86)	0.016
Left sided colitisa	1.32 (0.78–2.23)	0.305
Extensive colitisa	1.56 (0.90–2.70)	0.111	(−)
Proximal extension	1.75 (0.87–3.52)	0.118	(−)
Continuous active disease	10.7 (6.51–17.4)	<0.001	7.05 (4.23–11.77)	<0.001
Systemic steroids	3.52 (1.86–6.66)	<0.001	3.25 (1.37–7.71)	0.008
Topical steroids	1.41 (0.84–2.39)	0.198	(−)
5-ASA treatment	0.28 (0.12–0.65)	0.003	(−)
Immunomodulators	2.63 (1.51–4.58)	0.001	(−)
Cyclosporine	4.61 (2.84–7.48)	<0.001	2.80 (1.66–4.72)	<0.001
Biologicals	2.98 (1.86–4.78)	<0.001	(−)

^aDisease extent at diagnosis.

(−) Indicates a not significant variable that was excluded from multivariable analysis after forward stepwise regression.

UC: ulcerative colitis; EIM: extra-intestinal manifestation, PSC: primary sclerosing cholangitis; 5-ASA: 5-aminosalicylic acid; HR: hazard ratio; CI: confidence interval.

Discussion

In this historical cohort of incident UC cases at a single centre in the Netherlands, the disease extended towards the proximal colon in one-third of the patients with initial distal colitis during 10 years of follow-up. However, this disease extension was not associated with a higher incidence of colectomy. Conversely, continuous disease activity, use of systemic glucocorticosteroids and cyclosporine treatment were independent risk factors for colectomy. Furthermore, the cumulative colectomy rates remained rather high and did not change over time despite the introduction of biologicals.

Over the past few decades, studies have shown variable rates of disease location at diagnosis in UC. Studies published in the 1990s demonstrated that the majority of patients were diagnosed with either proctitis or extensive colitis and only a minority with left-sided colitis.^4,5 It is, however, questionable whether this is a true reflection of the disease distribution since there was no validated classification system such as the Montreal Classification at that time.⁹ Furthermore, it may have been more difficult to identify the most proximal extent of inflammation with less-advanced imaging techniques and somewhat more restricted access to endoscopy. Another explanation could be that disease behaviour has changed over time due to different treatment policies. In this study, we found that left-sided disease was most prevalent at diagnosis and extensive disease at the end of follow-up, which parallels the findings in more recent studies.^2,6

So far, limited data are available about UC disease behaviour patterns by describing and classifying the frequency and duration of disease relapses during a patient’s disease history. The IBSEN study prospectively evaluated disease behaviour in 454 UC patients with a follow-up of five years, based on the patient’s experience and a colonoscopy at diagnosis and five years later.¹⁰ The study showed that 59% of the patients experienced a decline in the severity of intestinal symptoms, 22% had a relapse-free disease course and 9% had chronic active disease. The present study showed a similar outcome based on objective measures such as the number of relapses and treatment initiation or dose escalation. In addition, we demonstrated that patients with chronic continuous disease have a two-fold increased risk for proximal disease extension and an eight-fold increased risk for colectomy. Therefore, it is of importance to identify the dynamic nature of the disease for prognosis and monitoring, which could be classified as shown in Figure 1.

This study showed that approximately one-third of the patients with distal colitis had progression to left-sided or extensive colitis during 10 years of follow-up. This rate of proximal disease extension is comparable to recently published studies by Solberg et al. and Vester-Andersen et al.^2,6 Interestingly, Solberg et al. showed that patients with proximal disease extension had a higher colectomy rate (28%) compared to those with extensive colitis at diagnosis (19%). This may reflect a more aggressive phenotype in these patients. In the present study, younger patients and patients with continuous active disease appeared to be the only independent predictors for proximal disease extension. However, we were not able to find a significant association between proximal disease extension and colectomy.

Over the past few decades, the published colectomy rates in UC have been declining.¹¹ Earlier studies reported colectomy rates as high as 25% at 10 years.^4,5 A recent Scandinavian study showed a nine-year cumulative colectomy rate of 14.5% in the early 1980s that decreased to 9.1% after the year 2003.¹² A large European multicentre study showed an overall 10-year cumulative colectomy rate of 8.7% even though one of the collaborating centres (the Copenhagen centre) had a much higher colectomy rate (25.7% at 10 years) than the other participating centres.¹³ The present study, however, showed a cumulative risk of colectomy of 17.2% at 10 years, which was more comparable to the data from Copenhagen. Of course, the high colectomy rate can be explained by significant referral bias. Patients with more severe and refractory UC are often referred to an academic institution. In line with a previous Dutch population-based study,¹⁴ our study showed that the cumulative colectomy rate did not decrease over time. However, the median time until colectomy was longer than previously described (4.5 years versus approximately two years).¹⁰ This finding suggests that the advent of biologicals may not prevent but rather delay colectomy in patients with more severe disease.

An important limitation of this study is the retrospective design, which is always prone to missing data. Colonoscopy was performed only on indication and therefore unavailable in some of the patients. Furthermore, some patients have been lost to follow-up or referred back to a primary/secondary care setting. Therefore, there may have been an overrepresentation of patients with more severe disease. The strength of the present study is that it is the first cohort in which immunomodulators and biologicals were commonly used. Furthermore, this study included a large cohort with a long follow-up period that focused solely on UC patients.

In conclusion, in this cohort we found that one-third of the patients with UC have inflammation that extends more proximally over time. Younger age at diagnosis and continuous active disease were independent risk factors for proximal disease extension. However, proximal disease extension was not a risk factor for colectomy, but this risk was rather associated with continuously active disease, systemic steroids and cyclosporine treatment. The cumulative colectomy rates have remained high over time despite the introduction of biologicals.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Declaration of conflicting interests

Saloomeh Sahami, Kadere Konté, Christianne J. Buskens, Pieter J. Tanis and Willem A. Bemelman have nothing to declare.

Mark Löwenberg declares the following: speaking and teaching: Abbvie, Dr Falk, Ferring Pharmaceuticals, Merck Sharp & Dohme and Tramedico; and research grants: AbbVie and Achmea Healthcare.

Cyriel Ponsioen declares the following: advisory committees or review panels: Abbvie Laboratories, Glaxo Smith Kline, Takeda; and grant/research support: Abbvie Laboratories, Schering-Plough Corp., Falk Pharma, and Tramedico.

Gijs van den Brink declares the following: consulting fees: Abbott laboratories; lecture fees: Abbott Laboratories, Merck Sharp & Dohme and Ferring Pharmaceuticals; and research grants: Abbott Laboratories, Crucell and Ferring Pharmaceuticals.

Geert R. D’Haens declares the following: advisory committees, consulting or review panels: Amgen, Jansen Biologics, Abbvie, Boehringer Ingelheim, Cellgene, Cosmo, Galapagos, Versant, AM Pharma, Ferring, TEVA Pharmaceuticals, Glaxo Smith Kline, Shire Pharmaceuticals Inc, Mitsubishi Tanabe Pharma, Novo Nordisk A/S, Pfizer Inc, Receptos, MSD, and UCB; grant/research support: Dr Falk Pharma, Abbvie, MSD, Jansen Biologics, Takeda, Given Imaging, and Photopill; and speaking and teaching: MSD, UCB, Abbvie, Ferring Pharmaceuticals, Takeda, Shire, Norgine, Giuliani, and Vifor.