Randomised clinical trial: mesalazine (Salofalk granules) for uncomplicated diverticular disease of the colon – a placebo-controlled study

Authors


Correspondence to:

Prof. W. Kruis, Evangelisches Krankenhaus Kalk, Buchforststrasse 2, 51103, Köln, Germany.

E-mail: kruis@evkk.de

Summary

Background

Robust evidence regarding medical intervention for symptomatic uncomplicated colonic diverticular disease (DD) is sparse.

Aim

To investigate mesalazine (Salofalk granules) in this setting.

Methods

In a double-blind, placebo-controlled, multicentre, 6-week trial, patients were randomised to mesalazine 1000 mg three times daily or placebo. Primary efficacy endpoint was change in lower abdominal pain to week 4 (baseline defined using pain score from 7 days pre-treatment).

Results

Median change in lower abdominal pain with mesalazine vs. placebo was −37 (= 56) vs. −33 (= 61) [P = 0.374; 95% CI (−11; 4)] in the intent-to-treat (ITT) population, and −41 (= 40) vs. −33 (= 51) [P = 0.053; 95% CI (−18; 0)] in the per-protocol (PP) population, i.e. the primary endpoint was not significantly different. Post hoc adjustment for confounding factors (‘baseline pain intensity’, ‘baseline symptom score (Brodribb)’, and ‘localisation of diverticula in the descending colon’) resulted in P = 0.111 [ITT, 95% CI (−15.4; 1.6)] and P = 0.005 [PP, 95% CI (−19.7; −3.5)]. Between-group differences increased using pain score on day 1 as baseline, and reached significance for the PP population [mesalazine −42, placebo −26, P = 0.010; 95% CI (−25; −3)]. Median change in combined symptom score from baseline to week 4 was 257 mm with mesalazine vs. 198 mm with placebo [P = 0.064; 95% CI (−3; 105)]. More placebo-treated patients received analgesic/spasmolytic concomitant medication (34.4% vs. mesalazine 21.4%), indicating improved pain relief with mesalazine (P = 0.119). Safety was comparable.

Conclusions

A daily dose of 3.0 g mesalazine may relieve pain during a symptomatic flare of uncomplicated DD. In this, the first placebo-controlled double-blind trial in acute uncomplicated DD, mesalazine showed promising therapeutic efficacy.

Introduction

Diverticula of the colon become increasingly common with age.[1, 2] Colonic diverticulosis is estimated to affect fewer than 10% of individuals under 40 years of age, but rises to more than 50% of those aged 70 or over.[3-5] Most individuals with diverticulosis remain free of major symptoms, perhaps experiencing only mild and nonspecific symptoms such as cramping or bloating, but 10–25% will develop diverticulitis.[1, 5] Of these, approximately three-quarters will have uncomplicated, but painful, diverticular disease (DD) while a small proportion develops complications such as abscesses or perforations that require surgical intervention.[5]

The clinical presentation of uncomplicated DD varies, ranging from a single acute attack or intermittent abdominal pain (frequently mistaken for irritable bowel syndrome) to chronic abdominal pain and haematochezia.[2, 6] Although there is no supporting evidence from placebo-controlled trials, recommendations for standard management of acute episodes of uncomplicated DD include medical treatment with broad-spectrum antibiotics.[6, 7] Alternatively, there is a growing body of evidence supporting the use of the anti-inflammatory agent mesalazine for symptomatic relief in uncomplicated disease.[8] There is preliminary evidence to suggest the presence of low-grade mucosal inflammation in symptomatic uncomplicated DD,[9, 10] and increased levels of neuropeptides in mucosal biopsies from patients with symptomatic DD that may reflect previous inflammation.[11] Two randomised comparative studies have evaluated mesalazine therapy for uncomplicated DD.[12-14] These showed a benefit for mesalazine compared with rifaximin in terms of preventing symptomatic recurrence,[12, 13] and similar success in maintaining long-term remission compared with the probiotic Lactobacillus casei.[14] However, neither trial included a placebo treatment arm nor used a double-blind design, and the study endpoints were not validated.[12-14] Other researchers have shown that mesalazine in combination with rifaximin, followed by mesalazine monotherapy, is effective in maintaining symptom remission but the single-centre study included no comparative arm.[15] A recent Cochrane review of published trials of mesalazine and other aminosalicylic agents concluded that this class of agents may have a role in the treatment of colonic DD.[16]

However, in the absence of placebo-controlled, double-blind studies, robust evidence for the benefits of medical intervention in acute uncomplicated DD, including mesalazine, remains sparse.

Here, we report the first randomised, placebo-controlled, double-blind study undertaken to explore the therapeutic efficacy of mesalazine (Salofalk granules; Dr Falk Pharma GmbH, Freiburg, Germany) in patients experiencing an acute symptomatic attack of uncomplicated DD.

Methods

Study design and conduct

This was a double-blind, randomised, placebo-controlled, parallel-group trial performed at 17 centres in Germany during the period May 2002 to August 2004. The objective of the study was to investigate the efficacy and safety of mesalazine [Salofalk granules 1000 mg three times daily (t.d.s.); Dr Falk Pharma GmbH] vs. placebo in patients with lower abdominal pain as a symptom of uncomplicated DD. The study was conducted in accordance with the Declaration of Helsinki and the ICH Guidelines for Good Clinical Practice, following approval from the Independent Ethics Committee at each participating centre. Written informed consent was obtained from each participant. The ClinicalTrials.gov identifier is NCT01627262 (http://clinicaltrials.gov/).

Patient population

The study population comprised male and female patients with a diagnosis of acute, symptomatic, uncomplicated DD. Patients were eligible to enter the study if they met the following criteria: age 45–80 years; diagnosis of DD with acute pain without serious complications (e.g. peritonitis, abscess, fistula, visible blood on stool not originating from haemorrhoids, ileus, stenosis); lower abdominal pain of moderate or severe intensity on at least four of the previous 7 days before study inclusion; a minimum of four diverticula observed on endoscopy (at least flexible sigmoidoscopy) examination at baseline; and at least four of eight specified symptoms present for at least the previous 2 days before inclusion and still present at study inclusion (i.e. abdominal pain localised mainly in the lower left part of the abdomen; abdominal pain enhanced after meals; abdominal pain decreased after defaecation or wind; bloating; constipation defined as ≤2 defaecations/week; diarrhoea, defined as >3 loose stools per day; a sensation of incomplete evacuation after defaecation; and painful lower left abdomen at palpation). Patients were excluded if they had chronic inflammatory bowel disease (Crohn's disease or ulcerative colitis); fever (>38.0 °C by axillary measurement) or other signs of serious complications; a medical history of severe renal disease, defined as serum creatinine >1.5 mg/dL; known intolerance to the study medication; or a requirement for prohibited concomitant medication.

Treatment and concomitant medication

Randomisation was performed at baseline visit (day 1) via a computer-generated randomisation list that used randomly permuted blocks. The randomisation list was held by staff at the Contract Research Organisation (CRO) Pharmalog Institut für klinische Forschung GmbH, Munich, Germany, which was not involved in the study conduct. Block randomisation (block size, = 4) was performed using an automated randomisation software programme (RANCODE; IDV-Datenanalyse und Versuchsplanung, Gauting, Germany). Eligible patients were randomised in a 1:1 ratio to receive either mesalazine 1000 mg t.d.s. or placebo 1000 mg t.d.s. for 6 weeks. Investigators received sequentially numbered containers and were instructed to allocate patients consecutively in the order of their enrolment. The investigators were kept blinded about the sequence until the study end.

Mesalazine was administered as Salofalk granules (Dr Falk Pharma GmbH), which differ from other mesalazine formulations by combining both delayed- and extended-release mechanisms. First, mesalazine release is delayed until pH ≥6.0 due to an enteric, acid-resistant film coating resulting in a mean lag time of about 3 h.[17] Second, due to inner polymer matrix the release of the active ingredient is prolonged throughout the entire colon, reflected by a lower extent of systemic drug absorption as compared to tablets, which possess only an enteric coating but no inner polymer matrix.[17] The placebo granules were identical in size, form and appearance to the mesalazine granules and thus could not be distinguished by the investigators or patients from the active therapy. A daily mesalazine administration regimen was selected based on evidence that this is more effective than cyclic dosing for maintaining remission in recurrent symptomatic uncomplicated DD.[18] Treatment compliance was calculated as the ratio between the administered medication (as determined by returned medication) and the expected intake during the actual treatment period of each patient.

All patients were instructed to follow nutritional recommendations including consumption of meals rich in fibre and adequate intake of liquids (at least 2 L/day).

No concomitant administration of any other drugs for treatment of gastrointestinal tract disorders was permitted that could affect the results or interfere with the study medication, with the exception of short-acting spasmolytics (e.g. butylscopolaminiumbromide) and short-acting analgesics (e.g. paracetamol). Opioids were prohibited.

Evaluation

Study visits took place at baseline (day 1) and at weeks 2, 4 and 6. Patients completed a daily diary from day 1 onwards, which was checked at all visits. The diary included daily scoring of the most severe lower abdominal pain intensity experienced, which was to be completed by the patient at the end of each day using a 5-point scale (1: ‘no pain’, 2: ‘slight pain’, 3: ‘moderate pain’, 4: ‘strong pain’, 5: ‘most severe pain’), the number of daily stools, and the main stool consistency. The combined symptom score according to Brodribb was assessed as each visit, using the sum of seven visual analogue scales (100 mm VAS) for seven questions.[19] At the end of the study, a global assessment of efficacy using a 4-point scale was provided by the patient (1: ‘ineffective’, 2: ‘moderately effective - slight improvement of complaints’, 3: ‘effective - marked improvement in symptoms’, 4: ‘very effective – as good as no symptoms’). Vital signs were measured at each visit, as well as laboratory analysis of urine and serum samples. Laboratory assessments were performed locally, and were analysed with reference to the normal ranges applied at each centre to allow for between-centre variations. Compliance with study medication was checked by pill counting, and concomitant use of spasmolytics and analgesics was reported. Adverse events were recorded at all study visits.

Study endpoints

The primary endpoint was the change in intensity of lower abdominal pain during the first 4 weeks of treatment, defined as the cumulative difference of the daily pain intensity score from the first day of study medication intake (day 1) to week 4 (Sum of Pain Intensity Differences, SPID1–28), as documented in the patient diaries. The baseline value was defined as the median of the daily maximum pain intensities during the previous 7 days before the start of treatment, as assessed retrospectively by the investigators during questioning of patients at study inclusion, using the same scale as that used in the diary. In addition, a more stringent but exploratory secondary analysis was performed, in which baseline pain intensity was expressed by the actual value for pain intensity recorded in the patient diary on day 1 (i.e. the first day of study medication intake), to exclude any potential recall bias. The latter definition of baseline pain intensity was introduced by a protocol amendment during the clinical phase of the study, and thus was predefined prior to database lock and unblinding i.e. it formed part of the initial analysis plan. Other secondary endpoints included the number of patients with complete pain relief and the median time to complete pain relief based on patient diary entries, the change in the combined symptom score (at week 4 and week 6), the number of patients requiring spasmolytics and analgesics, and the global assessment of efficacy as assessed by the patient and investigator. Safety endpoints included adverse events, laboratory results and vital signs.

Statistics

Sample size was estimated based on the assumption that a 30% difference in the primary endpoint between the mesalazine and placebo groups was clinically relevant. A sample of 96 patients (48 patients per group) was required to provide 80% power (1-β) to demonstrate a 30% difference using a nominal significance level of α = 0.05 (two-sided). To counter uncertainties due to limited information for adequate planning, a sample size of 120 patients (60 patients per group) was prespecified in the protocol.

The study was planned with a two-stage Bauer-Köhne approach. The stopping boundaries were defined to ensure a type-1 error level of α = 0.05 (two-sided). Since the futility margin was exceeded for the primary endpoint at the interim analysis, confirmatory testing was stopped but the trial was continued to reach the prespecified sample size. Exploratory analyses were undertaken based on data from all randomised and treated patients.

The primary endpoint analysis and other efficacy assessments were performed based on the intent-to-treat (ITT) population, which included all randomised patients who had received at least one dose of study medication. Key efficacy measures were also analysed in the per-protocol (PP) population, defined as all ITT patients who adhered to the study protocol without relevant deviations or violations. Safety analyses were performed on the safety population, defined as all randomised patients who received at least one dose of the study medication and for whom any data were available following the first dose of study medication.

All primary and secondary endpoints are presented by treatment group using summary statistics. The last observation carried forward (LOCF) method was used for calculation of the difference in pain intensity scores from baseline to week 4 and the percentage of patients with complete pain relief at week 6. Comparability between groups was tested by the Chi-squared or the Wilcoxon–Mann–Whitney test. Ninety-five per cent confidence intervals (CIs) were applied using the Hodges-Lehmann estimator of the treatment differences (HL-difference) for the Wilcoxon test of exact CI of percentage differences for the fourfold table situation. For the analysis of covariance (ancova), the 95% CIs of the least square means were calculated. For categorical variables, the number and percentage of patients for each category were calculated for nonmissing data within each treatment group. All statistical tests were performed at the two-sided nominal level of an error probability (α) of 0.05. As the overall results were of an exploratory nature, no alpha-error adjustment due to multiple testing was performed.

Additional analyses were performed to examine differences in the primary endpoint between the ITT and PP populations. The influence of confounding baseline variables as well as the effect of the intake of spasmolytic or analgesic medication was evaluated. The primary endpoint was adjusted for the intake of analgesic/spasmolytic medication by increasing the pain score to the maximum value (5) for the specific day that these medications were taken (worst case scenario), and was analysed in an ancova model including the confounding baseline variables as additional effects or covariates.

Data collection, monitoring and all statistical analyses were performed by the CRO using sas statistical software, version 8.2 and 9.2 (Cary, NC, USA).

Results

A total of 123 patients were recruited and randomised, of whom 117 received at least one dose of study medication and formed the ITT and safety populations (56 mesalazine, 61 placebo). The PP population comprised 91 patients (mesalazine 40, placebo 51), with use of prohibited concomitant medication being the most frequent reason for protocol violation (six and eight patients respectively) (Figure 1). In total, 14 mesalazine patients and five placebo patients discontinued the study prematurely, such that the study was completed by 98 patients (83.8% of the ITT/safety populations).

Figure 1.

Patient disposition.

Patient demographics were similar between treatment groups (Table 1). Both treatment groups were comparable with respect to histological and sigmoidoscopic assessments of the diverticula (Table 1), but there was a nonsignificant trend to a lower proportion of patients with diverticula in the descending colon in the mesalazine group (19.6% vs. 34.4% in the placebo arm, P = 0.073). The combined symptom score at baseline was approximately 10% higher in the mesalazine group vs. the placebo cohort (365.8 ± 115.4 vs. 330.8 ± 109.8, P = 0.091). The mean score for the straining scale (pressing on defaecation) was significantly higher in the mesalazine arm (Table 1). At least one of three inflammatory parameters [erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) or leucocytes] was elevated in 37.5% (21/56) of patients in the mesalazine group and in 41.0% (25/61) of patients in the placebo group (P = 0.700).

Table 1. Patient demographics and baseline characteristics. Continuous variables are shown as mean (s.d.)
Mesalazine (N = 56)Placebo (N = 61) P value
  1. a

    Chi-squared test.

  2. b

    t-test.

  3. c

    Wilcoxon two-sample test (z-approximation).

  4. d

    Endoscopic findings.

Male gender, n (%)22 (39.3%)21 (34.4%)0.586a
Age (years)62.8 (8.6)61.9 (8.6)0.954b
Height (cm)167.8 (8.9)165.4 (8.1)0.218b
Weight (kg)76.3 (11.5)79.2 (15.0)0.481b
Smoking status, n (%)
Non-smoker45 (80.4%)43 (70.5%)0.175a
Ex-smoker 566 (10.7%)5 (8.2%)
Smoker5 (8.9%)13 (21.3%)
Alcohol consumption, n (%)
Never24 (42.9%)35 (57.4%)0.139a
Occasionally31 (55.4%)25 (41.0%)
Regular1 (1.8%)1 (1.6%)
Baseline elevated ESR, CRP, or leucocytes, n (%)21 (37.5%)25 (41.0%)0.700a
Baseline ESR (1 h; ≥20 mm), n (%)10 (17.9%)16 (26.2%)0.277a
Baseline elevated CRP, n (%)13 (23.2%)17 (27.9%)0.565a
Baseline elevated leucocytes, n (%)2 (3.6%)5 (8.2%)0.292a
Baseline intensity of lower abdominal pain (median of last 7 days prior to baseline, scale 1–5)3.4 (0.6)3.5 (0.6)0.878b
Baseline intensity of lower abdominal pain (day 1, scale 1–5)2.6 (1.1)2.8 (0.8)0.235c
Baseline Combined symptom score (mm VAS 0–100)366 (115.4)331 (109.8)0.091
Symptom scores at baseline (mm VAS 0–100)
Intensity of lower abdominal pain61.6 (18.8)60.8 (20.7)0.892b
Frequency of lower abdominal pain70.2 (21.6)70.0 (21.8)0.941b
Frequency of bloating/abdominal distension61.8 (29.2)58.5 (32.5)0.700b
Wind, flatulence52.6 (27.8)51.8 (28.4)0.883b
Pressing during defaecation41.9 (30.4)30.4 (29.9)0.039b
Sensation of incomplete evacuation after defaecation51.3 (32.8)40.7 (29.5)0.064b
Nausea26.5 (27.8)18.7 (24.9)0.153b
Number of diverticula/patient, n (%)
4–73 (5.4%)6 (9.8%)0.265a
8–1012 (21.4%)7 (11.5%)
>1041 (73.2%)48 (78.7%)
Presence of diverticula
Sigmoid colon56 (100.0%)60 (98.4%)0.336a
Descending colon11 (19.6%)21 (34.4%)0.073a
Transverse colon5 (8.9%)7 (11.5%)0.650a
Ascending colon5 (8.9%)3 (4.9%)0.391a
Visible inflammation at rim of diverticulad 8 (14.3%)9 (14.7%)0.994
Visible inflammation of peridiverticular mucosad 7 (12.5%)6 (9.8%)0.567

Study drug and concomitant medication

Compliance with treatment, calculated as a percentage of the scheduled dose [mean ± standard deviation (s.d.)], was 94.9 ± 14.1% and 99.5 ± 8.8% in the mesalazine and placebo groups respectively.

In total, 12 of 56 patients (21.4%) in the mesalazine group and 21 of 61 patients (34.4%) in the placebo arm used concomitant analgesics or spasmolytics (P = 0.119). The mean duration of either analgesic or spasmolytic therapy during the 6-week study was 0.93 days/patient in the mesalazine group vs. 1.75 days/patient in the placebo arm. For analgesic therapy, the corresponding values were 0.43 days/patient for mesalazine and 0.84 days/patient for placebo, for paracetamol the mean values were 0.18 days/patient and 0.23 days/patient, and for the antispasmodic hyoscine butylbromide the mean values were 0.32 days/patient and 0.77 days/patient respectively.

Efficacy

Primary endpoint

The primary endpoint, i.e. the sum of differences in lower abdominal pain intensity from baseline to week 4 calculated over the first 4 weeks of treatment (SPID1–28), was a median of −37 points (range: −95–25) under mesalazine and −33 score points (range: −78–24) under placebo, a difference that was not significant [P = 0.374; HL–difference −3.5; 95% CI (−11; 4); ITT population]. When the same analysis was repeated in the PP population, the between-group difference approached significance [mesalazine: median −41 (range −95 to −7); placebo: median −33 (range −78–24); P = 0.053; HL-difference −9; 95% CI (−18; 0)] (Figure 2a). As per protocol, the change in lower abdominal pain to week 4 was also calculated using the patient's pain intensity score on day 1 as baseline. Using this definition, the difference between groups was wider than for the primary endpoint, which used retrospective data to define the baseline value, and the between-group difference reached significance for the PP population [P = 0.010; HL-difference −14; 95% CI (−25; −3)] (Figure 2b).

Figure 2.

Cumulative difference of the daily pain intensity score from baseline to week 4 (Sum of Pain Intensity Differences, SPID1–28) for (a) baseline score defined as the median of the previous seven days prior to treatment and (b) baseline score defined on day 1. ITT, intent-to-treat; PP, per-protocol. Last observation carried forward (LOCF) method.

ancova analysis

A post hoc ancova analysis was performed to examine the differences in the primary endpoint between the ITT and PP populations and reveal further ‘confounders’ responsible for the higher P value in the ITT population. For this analysis, SPID data were adjusted for the intake of analgesic or spasmolytic medication by increasing the pain score to the maximum value (5) for the specific day that these medications were taken (worst case scenario). After adjustment for the confounder ‘baseline lower abdominal pain intensity (median from Day-7 to Day-1)’, ‘localisation of diverticula in the descending colon’ and ‘baseline combined symptom score (Brodribb)’ in the ancova model, the resulting P values for the between-group differences in the primary endpoint decreased strongly for the ITT population [P = 0.111; adjusted difference −6.9; 95% CI (−15.4; 1.6)] and the PP population [P = 0.005; adjusted difference −11.6; 95% CI (−19.7; −3.5)].

Secondary endpoints

Median time to complete pain relief was 8 days (95% CI (4; 17)] under mesalazine and 19 days [95% CI (10; 25)] under placebo in the ITT population (P = 0.105), a difference that was significant in the PP population: 8 days [95% CI (4; 16)] under mesalazine and 22 days [95% CI] under placebo (P = 0.013) (Figure 3a). By week 6, the proportion of patients with complete pain relief was 73.2% (41/56) and 65.6% (40/61) in the mesalazine and placebo groups respectively [P = 0.371; difference: 7.6%; 95% CI (−10.7%; 25.5%)]. Similar results were obtained in the PP population, with the difference between mesalazine [82.5% (33/40)] and placebo [64.7% (33/51)] approaching significance [P = 0.059; difference 17.8%; 95% CI (−2.9%; 37.5%)] (Figure 3b). Improvement of lower abdominal pain by week 4 was reported by 71.4% of patients in the mesalazine group (40/56) and 67.2% (41/61) of the placebo patients using baseline values derived from retrospective data from the 7 days prior to study entry [P = 0.622; difference 4.2%; 95% CI (−14.1%; 22.2%)]. Within each group, there was a significant mean reduction of pain score from baseline to week 4 under mesalazine (1.4 ± 1.26 points, P < 0.001) and placebo (1.1 ± 1.34 points, P < 0.001).

Figure 3.

Complete pain relief: (a) Median time to complete pain relief and (b) number (%) of patients reporting complete pain relief at week 6. ITT, intent-to-treat; PP, per-protocol.

The median changes in combined symptom score from baseline using the 100 mm visual analogue scoring system approached significance between the mesalazine and the placebo group at both week 4 and week 6 (see Table 2). All seven individual scores showed numerically greater improvements with mesalazine vs. placebo, the largest of which was observed for intensity of lower abdominal pain (47.8 ± 24.3 mm vs. 40.8 ± 22.9 mm respectively [P = 0.063; HL-difference 9 mm; 95% CI (−1 mm; 19 mm)].

Table 2. Secondary efficacy parameters (ITT population)
 Mesalazine (N = 56)Placebo (N = 61)P-valuea95% CI of differencesd
  1. a

    P-value for difference between groups.

  2. b

    According to Brodribb.[19]

  3. c

    Cochran-Armitage trend test.

  4. d

    Hodges-Lehmann Estimator.

Change in combined symptom score (mm VAS 1–100)b [median (range)]
Baseline to week 4256.5 (−17–614)198.0 (−134–494)0.064−3–105
Baseline to week 6262.5 (8–589)205.5 (−85–496)0.065−3–96
Global efficacy assessment by investigator, n (%)
‘Very effective’16 (28.6%)14 (24.0%)0.053c
‘Effective’19 (33.9%)18 (29.5%)  
‘Moderately effective’6 (10.7%)19 (31.1%)
‘Ineffective’4 (7.1%)7 (11.5%)  
Missing11 (19.6%)3 (4.9%)
Global efficacy assessment by patient, n (%)
‘Very effective’18 (32.1%)15 (24.6%)0.040c
‘Effective’17 (30.4%)16 (26.2%)  
‘Moderately effective’6 (10.7%)21 (34.4%)
‘Ineffective’4 (7.1%)6 (9.8%)  
Missing11 (19.6%)3 (4.9%)

For 35 of 56 patients (62.5%) treated with mesalazine, the investigator rated the efficacy of the treatment as ‘effective’ or ‘very effective’ compared with 32 of 61 patients (52.5%) in the placebo group (Table 2; Cochran-Armitage trend test P = 0.053). The patients' global self-rating of treatment efficacy corresponded well with the investigators' assessment: 35/56 mesalazine-treated participants (62.5%) and 31/62 placebo-treated patients (50.8%) rated the treatment as ‘effective’ or ‘very effective’ (Table 2; Cochran-Armitage trend test: P = 0.040).

Safety

In total, 18 adverse events were reported in 13 patients (23.2%) randomised to mesalazine and 17 adverse events occurred in 11 patients (18.0%) in the placebo group. The most frequently reported adverse events were headache [1 mesalazine patient (1.8%), 3 placebo patients (4.9%)] and diarrhoea [no mesalazine patients, 2 placebo patients (3.3%)]. No other adverse event was reported in more than one patient. The majority of events were of mild (= 14) or moderate (= 17) intensity and subsided spontaneously or after symptomatic treatment. Two adverse events (fullness and severe, persistent epigastric pain) in one placebo-treated patient were attributed to study drug, and none in the mesalazine arm. Two serious adverse events (allergic reaction to a concomitant medication and psychosomatic disease) were reported in two placebo-treated patients. No serious adverse events occurred in mesalazine-treated patients. There were no deaths during the study. Four patients in the mesalazine group and two patients in the placebo group discontinued the study prematurely due to adverse events; none of these events was attributed to the study drug. The determination of vital signs, and measurement of a wide range of biochemical and haematological parameters, did not reveal any significant changes within either group during the study.

Discussion

In this, the first placebo-controlled trial of mesalazine for acute uncomplicated DD, mesalazine treatment may be superior to placebo for alleviating lower abdominal pain and other efficacy outcomes. Analyses consistently showed significant differences in favour of mesalazine vs. placebo in the PP but not the ITT population. Several factors may have contributed to this inconsistency. For example, the use of concomitant analgesic/spasmolytic medication was numerically more frequent in the placebo arm, which may have influenced the proportion of patients reporting an improvement in pain and diminished differences in treatment effects between the groups. To account for this, SPID data were adjusted for the intake of analgesic or spasmolytic medication, followed by an ancova analysis to identify further confounding factors. Three relevant factors were identified: ‘localisation of diverticula in the descending colon’, ‘baseline intensity of lower abdominal pain (median from Day-7 to Day-1)’, and ‘baseline combined symptom score (Brodribb)’. After adjustment for these factors in the ancova model, the P value for therapeutic efficacy mesalazine vs. placebo decreased strongly from P = 0.374 to P = 0.111 in the ITT population. Consistent with this, there was a nonsignificant trend to a higher proportion of patients rating mesalazine as effective or very effective compared with the placebo group.

Very few studies, with inconsistent results, have investigated the influence of localisation and number of diverticula on the course of DD, and a potential confounding effect of baseline differences in these factors cannot be ruled out.[20] To rule out such a potential confounding effect, adjustment for varying distribution between groups is meaningful. Non-adherence clearly influences the outcomes of treatments[21] and probably contributed to the differences in treatment effect between ITT and PP populations.

An important obstacle to the appropriate evaluation of medical therapies for uncomplicated DD is the absence of established diagnostic criteria and markers of severity and, thus, recognised efficacy endpoints. Diagnosis of DD based purely on clinical symptoms is challenging, particularly with regard to excluding a diagnosis of irritable bowel syndrome. Leucocytosis and fever may clearly indicate diverticulitis,[22] but poor sensitivity and limited diagnostic accuracy of laboratory parameters such as ESR, CRP or leucocytosis have been described, particularly in uncomplicated diverticulitis.[23, 24] Therefore, in patients with elevated levels of ESR, CRP or leucocytes, the term ‘diverticular disease’ was introduced.

Colonoscopy may be essential for the diagnosis of comorbidity and complications in patients with diverticulosis,[25] but because DD occurs mainly within or outside the herniated mucosa, intra-luminal investigation by colonoscopy cannot measure disease activity. Although ultrasound is the gold standard evaluation technique,[26] computer tomography (CT) scanning remains popular. Both methods can detect complications such as abscesses, fistulae and stenosis but the prognostic value of CT scanning for severity of DD is limited.[27] Thus, clinical symptoms remain a valid endpoint measure. Many symptoms can contribute to the clinical presentation, but abdominal pain is considered a hallmark of the disease,[28] and we therefore focussed on pain as our primary outcome measure.

Except in rare cases, colonic diverticula are histologically pseudodiverticula, consisting of herniation of the mucosa and submucosa through the muscular wall of the colon. Diverticulitis is thus essentially an inflammatory reaction of the mucosa, justifying a therapeutic approach similar to that employed for other inflammatory diseases of the colonic mucosa such as ulcerative colitis. Mesalazine exerts multiple anti-inflammatory effects,[29] and for many years has been the mainstay of medical management for mildly to moderately active ulcerative colitis,[30, 31] making it a promising candidate for the treatment of other intestinal inflammatory disease such as DD. Currently, broad-spectrum antibiotics are routinely used for uncomplicated DD,[6, 7] although robust supportive data are scarce.[32] A retrospective analysis in 311 patients with acute colonic diverticulitis admitted to a surgical hospital concluded that use of antibiotics did not affect the rate of recurrence or the need for subsequent surgery.[33] A recent randomised trial in acute uncomplicated diverticulitis could not demonstrate any therapeutic advantages of antibiotics.[34] In view of the risk of serious complications, such as antibiotic-associated diarrhoea and infection with Clostridium difficile, alternative treatments to antibiotics are urgently needed. In the current study, mesalazine was very well tolerated with a safety profile similar to placebo, consistent with long-term experience in other indications, which would tend to support the use of mesalazine in this setting.

Our study has some limitations which should be taken into account. Adequate planning of the study was difficult and sample size calculation was hampered due to a lack of efficacy data from previous placebo-controlled studies in this indication. Therefore, this study may have been underpowered. Concomitant analgesic or other medication appears to be indispensable in this setting, but with hindsight the type and dose of drugs should have been standardised and considered a priori within the outcomes analysis. Nevertheless, in view of the lack of established study designs in this field, the study provides important planning data for further confirmative trials in uncomplicated DD.

In conclusion, a daily dose of 3.0 g mesalazine was safe and may achieve relief of pain in patients with a painful flare of uncomplicated DD. In this, the first placebo-controlled double-blind trial in acute uncomplicated DD, mesalazine showed promising therapeutic efficacy and the study results provide relevant data for the design of further controlled trials.

Authorship

Guarantor of the article: Wolfgang Kruis.

Author contributions: WK, EM, MS and OM performed the research. WK, RG and RM analysed the data. All authors approved the final version of the manuscript.

Acknowledgements

Declaration of personal interests: WK has served as a speaker, a consultant and/or an advisory board member for Dr Falk Pharma GmbH. RG and RM are employees of Dr Falk Pharma GmbH.

Declaration of funding interests: This study was funded in full by Dr Falk Pharma GmbH, Freiburg, Germany. The preparation of this paper was undertaken by WK, with writing support provided by C Dunstall, a freelance writer funded by Dr Falk Pharma GmbH.

Appendix A: The German SAG-20 Study Group

Dr med. Jochen Arnold, Regensburg; PD Dr med. Hans Bellmann, Zweibrücken; Dr med. Petra Berndt, Berlin; Dr med. Bernd Bokemeyer, Minden; Dr med. Axel Dettmer, München; Dr med. Albert Eimiller, München; Dr med Thomas Eisenbach, Leverkusen; Dr med. Rolf Fink, Freising; Dipl. med. Wolfgang Grosse, Potsdam; Dr med. Dietrich Hüppe, Herne; Dr med. Heinz-Jochen Kramm, Berlin; Prof. Dr med. Wolfgang Kruis, Köln (Co-ordinating investigator); Dr med. Thomas Krummenerl, Münster; Dr med. Eberhard Meier, Amberg; Dr med. Oliver Mickisch, Mannheim; Dr med. Stefan Pape, Paderborn; Dr med. Werner Resch, Landshut; Dr med. Michael Römer, München; Dr med. Thomas Schädlich, Ellefeld; PD Dr med. Werner Schneider, Halle; Dr med. Michael Schumacher, Wolmirstedt; Dr med. Maria–E. von Gynz-Rekowski, Berlin; Dr med. Elmar Zehnter, Dortmund; Dr med. Jürgen Zeus, Erlangen.

Ancillary