Meta-analysis: the utility and safety of heparin in the treatment of active ulcerative colitis

Authors

  • J. SHEN,

    1. Department of Gastroenterology, Renji Hospital, Shanghai Jiao Tong University, School of Medicine, Shanghai Institute of Digestive Disease, Shanghai, China
    Search for more papers by this author
  • Z. H. RAN,

    1. Department of Gastroenterology, Renji Hospital, Shanghai Jiao Tong University, School of Medicine, Shanghai Institute of Digestive Disease, Shanghai, China
    Search for more papers by this author
  • J. L. TONG,

    1. Department of Gastroenterology, Renji Hospital, Shanghai Jiao Tong University, School of Medicine, Shanghai Institute of Digestive Disease, Shanghai, China
    Search for more papers by this author
  • S. D. XIAO

    1. Department of Gastroenterology, Renji Hospital, Shanghai Jiao Tong University, School of Medicine, Shanghai Institute of Digestive Disease, Shanghai, China
    Search for more papers by this author

Dr Z. H. Ran, Department of Gastroenterology, Renji Hospital, Shanghai Institute of Digestive Disease, Shanghai 200001, China.
E-mail: z-ran@online.sh.cn

Abstract

Summary

Background

The use of heparin for the treatment of ulcerative colitis has been evaluated in several open and controlled trials, with varying outcomes.

Aim

To evaluate the efficacy and safety of heparin as supplemental therapy compared with conventional therapy in patients with ulcerative colitis.

Methods

All randomized trials comparing heparin supplementation to conventional therapy were included from electronic databases. Statistical analysis was performed with review manager 4.2.8 (The Cochrane Collaboration, Oxford, UK). Sub-analysis and sensitivity analysis were also performed.

Results

Eight randomized-controlled trials, investigating a total of 454 participants, met the inclusion criteria. The odds ratio (OR) for the efficacy of heparin supplementation vs. conventional therapy was 0.78 (95% CI = 0.50–1.21). Few serious adverse events were observed. The OR for the efficacy of unfractionated heparin and low-molecular-weight heparin vs. conventional therapy was 0.26 (95% CI = 0.07–0.93) and 0.92 (95% CI = 0.57–1.47), respectively. The OR for the efficacy of heparin vs. conventional therapy with placebo was 0.87 (95% CI = 0.53–1.44).

Conclusions

Our meta-analysis suggests that administration of heparin in patients with ulcerative colitis is safe, but no additive benefit over conventional therapy is indicated.

Introduction

The pathophysiological mechanisms responsible for the development of ulcerative colitis (UC) remain unclear. The conventional therapy with aminosalicylates, corticosteroids and immunosuppressive agents is still widely used, despite the fact that a significant proportion of patients is refractory to therapy or requires surgery.

Since 1980s, there has been an interest in the use of heparin as a treatment for UC. There is a growing body of evidence supporting the use of heparin in the treatment of UC.1–5 Heparins reverse the endothelial dysfunction which may be important in initiating and perpetuating the UC.6 Heparin has anticoagulant and anti-inflammatory properties. As part of the inflammatory response in inflammatory bowel disease, patients are at increased risk of thrombosis and platelets activation, which lead to the bowel symptoms in UC.7

Many clinical trials including uncontrolled and controlled trials have evaluated efficacy of both unfractionated heparin (UH) and low-molecular-weight heparin (LMWH) in ameliorating the intestinal inflammation. A number of small studies have explored the encouraging results of heparin in the treatment of UC. These trials have all shown a benefit to the use of heparin in UC and for the most part with little associated toxicity. A UH has been demonstrated encouraging efficacy in the treatment of UC since 1980s. In some open-label trials, patients even unresponsive to corticosteroid therapy treated with UH achieved clinical improvement. Low-molecular-weight heparin such as fragmin, nadroparine, enoxaparin, etc. also led to clinical, endoscopic and histologic improvement in some uncontrolled trials.8–10

In contrast, randomized-controlled trials (RCT) comparing heparin with conventional treatment including aminosalicylates, steroids or azathioprine showed that neither UH nor LMWH supplement groups had significant improvement in clinical, endoscopic, histologic or serologic assessment. It remains possible that heparin is effective in UC only when given with concomitant conventional therapy.

The aim of this meta-analysis was to evaluate the efficacy and safety of heparin therapy compared with conventional therapy in patients with inflammatory bowel disease.

Methods

Search strategy

A literature search of MEDLINE (1966 to January 2007), EMBASE (1980 to January 2007), the Cochrane Controlled Trials Register (1st Quarter, 2007), OVID (1950 to January 2007), BIOSIS (1996 to December 2006) and the Chinese Biomedical Database (1981 to December 2006) was used to identify the comparative studies of heparin in inflammatory bowel disease. The following keywords were used in combinations of the search: ‘heparin’, ‘inflammatory bowel disease’ and ‘UC’. A comprehensive search of reference lists of all review articles and original studies retrieved by this method was performed to identify additional reports. Furthermore, we hand searched abstracts of major gastroenterological meetings, such as the Digestive Disease Week of the American Gastroenterological Association and the World Congress of Gastroenterology. No language restrictions were made. Authors of some identified trials were asked whether they knew of additional studies, including unpublished randomized ones.

Inclusion criteria and exclusion criteria

The selection criteria were as follows.

Only RCTs were included. Studies were not included if they did not provide details on the patient selection, allocation, study design, outcome and measurement methods. Studies in abstract form or meeting report, without publication of the full paper, were also included in the analysis.

Only adult patients were included in the analysis.

Studies included at least two branches: conventional group received aminosalicylates, steroids or/and azathioprine with/without placebo; heparin group received UH or LMWH, or heparin group plus the same conventional treatment.

Papers were included if they provided information on at least one outcome parameter as follows: the response rate, the complete response (CR), the partial response (PR), the clinical disease activity index, the endoscopic assessment, the histologic assessment, the adverse events and the withdrawals.

Furthermore, articles published in English and other languages were also included if a translation was provided by the authors.

Data extraction

All the data were tabulated. Standardized data abstraction sheets were prepared. Data were exacted for author and year, location of trials, trail design, disease of patients, number of enrolled subjects, preparations and dose of administration, and study quality; meanwhile, key outcome data, such as response rates, rectal bleeding, toxic megacolon, nausea and surgery, were abstracted from the selected studies. All papers were examined independently for eligibility by two reviewers (Shen and Tong). Disagreements were resolved by consulting a third reviewer (Ran). Quality was assessed by using the Jadad score system based on the three items: randomization, double blinding and description of withdrawals and dropouts, to generate scores from 0 to 5. The selected studies were scored independently by two investigators (Shen and Tong), and if there were disagreements, then they discussed or consulted a third reviewer to obtain the final scores (Ran). We considered that they were low quality when their scores were lesser than three. When two or more publications from one institution appeared to review the same patients, only the most recent study results were included.

Statistical analysis

Dichotomous data were tested by calculating the rate difference with their 95% CI. The outcome measure examined was the odds ratios (OR) of response rates with heparin vs. conventional therapy based on a fixed-effects model using the methods of DerSimonian and Laird by an intention-to-treat. Heterogeneity between the studies was assessed by chi-squared test. Statistical significance for the test of heterogeneity was set at 0.10. If significant heterogeneity exists, it would be inappropriate to combine the data for further analysis using a fixed-effects model, while the random model was used for calculations.

Sub-analyses for the meta-analysis were planned depending on heparin preparations and study designs. To evaluate the stability of the results of this meta-analysis, we performed sensitivity analyses by estimating the ORs in the absence of one or more studies. Statistical analysis was performed with the software revman 4.2.8 from The Cochrane Collaboration (Oxford, UK).

Analyses of some variables were not possible because of the lack of uniformity of data presentation. For these studies, uniform data were summarized in evidence tables.

Results

Description of the studies

We reviewed 593 citations and abstracts obtained from our computerized literature searches and by hand-searching. Eight RCTs11–18 comparing heparin therapies with only conventional therapies for UC were identified to meet the inclusion criteria. Abstracts and full texts of remaining eight RCTs were retrieved for further considerations. Two articles17, 18 did not mention response rates, but mentioned the adverse events. Finally, six RCTs11–16 contained data for response rates. Among these six articles, one included not only patients with UC, but also three patients with Crohn’s disease; one compared heparin without concomitant treatment to conventional treatment. The data of Ang et al. were corrected before analyses so that only data on patients with UC were used. The characteristics of eight trials included in meta-analysis are summarized in Table 1, including quality score.

Table 1.   Characteristics of included studies in this meta-analysis
Author and yearLocationTrial designPatientsTotal (heparin/control)Heparin (type/dose)Conventional treatment (type/dose)Jadad scores
  1. RCT, randomized controlled trial; UC, ulcerative colitis; UH, unfractionated heparin; i.v., intravenous; s.c., subcutaneous; p.o, oral administration; b.d., twice daily; t.d.s., three times a day.

Panes et al.11 2000SpainMulticenter RCTUC (moderate and severe)25 (12/13)UH/31 456 ± 290 IU/day i.v. 10 daysMethylprednisolone/0.75–1 mg/kg/ day 10 days5
Ang et al.12 2000IrelandSingle-center RCTUC and CD (severe)20 (8/12)UH/25 000–45 000 IU/day i.v. 5 days 10 000 IU/b.d. 2 weeks 5000 IU/b.d. 3 weeks5-aminosalicylate/3–4 g/day 6 weeks Hydrocortisone 200 mg/day i.v. 5 days After 5 days of i.v. corticosteroids, patients were transferred to oral prednisolone 40 mg/day .The dose was reduced by 5 mg each week.3
Bloom et al.13 2003UK Denmark Canada IrelandMulticenter RCT double-blind, placebo-controlledUC (mild to moderate)100 (48/52)Tinzaparin/175 anti-Xa IU/kg/day s.c. 2 weeks Followed by 4500 anti-Xa IU/kg/day s.c. 4 weeksSalicylates/not specified3
Korzenik et al.14 2003USAMulticenter RCT double-blind, placebo-controlledUC (active)138 (92/46)Deligoparin/75 or 125 mg/day s.c. 6 weeksAminosalicylates steroids Azathioprine/not specified5
Zezos et al.15 2006GreeceSingle-center RCTUC (active)34 (16/18)Enoxaparin/100 Anti-Xa IU/kg/day s.c. 12 weeksAminosalicylates/2–4 g/day p.o. Steroids (prednisolone 50–75 mg/day i.v. or methyl-prednisolone 32–48 mg/day p.o. After initial response, i.v. corticosteroids were switched to oral, while methyl-prednisolone was tapered off during a period of 12 weeks.3
de Bievre et al.16 2007The NetherlandsMulticenter RCT double-blind, placebo-controlledUC (mild or moderate)29 (15/14)Reviparin/3.436 IU/b.d. s.c. 8 weeksSalazopyrine or mesalazine or olsalazine/2–3 g/day 8 weeks5
Korzenik et al.17 1999USAMulticenter RCTUC (moderate and severe)70 (56/14)UH/10 000U/b.d. and t.d.s. s.c. 6 weeksMesalazine steroids azahioprine/not specified5
Torkvist et al.18 2001SwedenMulticenter RCT placebo-controlledUC (mild to moderate)41 (21/20)Dalteparin/5000 IU s.c. b.d. 8 weeksAminosalicylates/not specified; steroids/not specified5

Response rates

Six studies described response rates (excluded two studies containing no data for response rates) were selected for the meta-analysis. Response rates included remission or CR to therapy and PR. Pooled response rates achieved in 80 of 189 patients including CR and PR were 48% (95% CI = 17.16–79.77%) for patients with heparin supplementation. Pooled response rates were 56% (95% CI = 23.39–89.47%) in 78 of 154 patients without heparin supplementation by intention-to-treat analysis. The OR for the efficacy of heparin supplementation vs. conventional therapy was 0.78 (95% CI = 0.50–1.21) (Figure 1.).

Figure 1.

 Odds ratio for response rates in heparin compared with conventional therapy in all pooled trials (n/N = patients with response/total patients).

Summary of effects

The effects of heparin for each study compared with conventional therapy are summarized in Table 2. No significant difference in any outcome measured such as clinical, endoscopic, histologic and serologic evaluation between either heparin or conventional groups.13–18

Table 2.   Summary of effects of heparin as adjuvant therapy compared with conventional therapy
Author and yearGroupClinical scoresEndoscopic scoresHistologic scoresESRCRPP-value
  1. ESR, erythrocyte sedimentation rate; CRP, C-reactive protein; H, heparin group; C, conventional treatment group; NR, not reported; NSI, no significant improvement; NSD, no significant difference; NS, not specified.

Korzenik et al.17 1999HNRNRNRNRNRNR
CNRNRNRNRNR
Panes et al.11 2000HNSINRNRNRNRNR
CImprovedImprovedNRImprovedImproved
Ang et al.12 2000HImprovedImprovedImprovedNRImprovedNSD
CImprovedImprovedImprovedNRImproved
Torkvist et al.18 2001HNSNSNRNRNRNSD
CNSNSNRNRNR
Bloom et al.13 2003HNSINSINSINRNRNSD
CNSINSINSINRNR
Korzenik et al.14 2003HNSNSNRNRNRNSD
CNSNSNRNRNR
Zezos et al.15 2006HImprovedImprovedImprovedNSINSINSD
CImprovedImprovedImprovedImprovedNSI
de Bievre et al.16 2007HNSINSINSINRNRNSD
CNSINSINSINRNR

Clinical score  Clinical score for UC was assessed in seven studies.12–18 Six of those studies found no significant difference in the clinical improvement between subjects treated with heparin as the adjuvant therapy and subjects treated with conventional drugs.13–18 Bloom et al.13 and Korzenik et al.17 reported statistically significant clinical improvement in both heparin adjuvant and conventional therapy. However, Zezos et al.15 and Torkvist et al.18 reported no clinical improvement in heparin group at the study endpoints. Joshua et al.14 and de Bievre et al18 showed no significant differences between heparin and conventional groups.

Endoscopic score  Endoscopic score was reported in seven studies.12–18 Six of those studies reported no significant difference between subjects treated with heparin as the adjuvant therapy and subjects treated with conventional drugs.13–18 Outcome of endoscopic improvement was similar to clinical score.

Histologic score  Histological score was reported in four studies.13, 15, 17, 18 No significant difference showed in any of the studies between heparin groups and conventional groups. Two reported significant histological improvement in both heparin adjuvant and conventional therapy13, 17 and two reported no significant improvement at the study endpoints.15, 18

ESR and CRP Erythrocyte sedimentation rate (ESR) was reported in two studies.12, 17 One reported significant ESR reduction only in conventional group.17 C-reactive protein (CRP) was reported in three studies.12, 13, 17 One studies reported statistically significant CRP reduction in both heparin adjuvant and conventional therapy.13 Another reported no significant CRP reduction at the study endpoint.17 No significant difference showed in any of the studies between heparin groups and conventional groups.

Adverse events

Few serious adverse events were observed in groups with or without heparin supplementation. The study by Korzenik et al.11 reported that three patients noted a transient increase in rectal bleeding, but was able to continue their involvement. The study by Panes et al.12 reported that three patients in the heparin group were withdrawn from the study because of increased rectal bleeding. One of these patients required urgent surgery. One of the patients in conventional group developed toxic megacolon and was managed with urgent surgery. In the study by Bloom et al.,15 only one patient, randomized to placebo, reported a significant rectal haemorrhage, and one in the same group reported nausea. Recently, the study by de Bievre et al.18 also reported three cases of nausea in heparin group and two in another. There was no significant difference in adverse events between the two study groups in any trials.

Sub-analysis

Sub-analysis for the meta-analysis was planned depending on the preparations before enrollment. We divided all eligible trials into UH subgroup (two trials) or LMWH subgroup (four trials). The OR for the efficacy of UH vs. conventional therapy was 0.26 (95% CI = 0.07–0.93). The OR for the efficacy of LMWH vs. conventional therapy was 0.92 (95% CI = 0.57–1.47) (Figure 2.).

Figure 2.

 Sub-analysis for response rates in heparin compared with conventional therapy based on the preparations (n/N = patients with response/total patients).

Based on the different study design, we also divided all eligible trials into ‘placebo controlled trials’ subgroup (three trials) or ‘controlled trials without placebo’ subgroup (three trails). The OR for placebo controlled trials was 0.87 (95% CI = 0.53–1.44). The OR for controlled trials without placebo was 0.55 (95% CI = 0.22–1.34) (Figure 3.).

Figure 3.

 Sub-analysis for response rates in heparin compared with conventional therapy based on the different study designs (n/N = patients with response/total patients).

Sensitivity analysis

We provided three sensitivity analyses depending on the properties of included studies. Firstly, we excluded the study by Ang et al.,12 in which subjects with both UC and Crohn’s disease were assessed, though we excluded the data of Crohn’s disease when we analysed overall response rates. In the five studies left,11, 13–16 only patients with UC were included. In the sensitivity analysis, response rates had no significant changes (OR 0.74; 95% CI 0.48–1.16) (Figure 4). Furthermore, the fixed-effects model was used because that the test for heterogeneity indicated the heterogeneity was acceptable (= 0.14). Secondly, we excluded the study by Panes et al.,11 in which heparin was used as monotherapy without concomitant conventional treatment. In the sensitivity analysis, response rates had no significant changes (OR 0.96; 95% CI 0.61–1.53) (Figure 5). Also, the fixed-effect model was also used because that the test for heterogeneity indicated the heterogeneity was acceptable (= 0.84). Finally, we excluded the study by Ang et al.12 and Panes et al.11 Only trials assessed patients with UC and comparing heparin as concomitant treatment with conventional therapy were left13–16. The fixed-effects model was also used because that the test for heterogeneity indicated the heterogeneity was acceptable (= 0.89). Response rates had no significant changes (OR 0.92; 95% CI 0.57–1.47) (Figure 6).

Figure 4.

 Sensitivity analysis for response rates in heparin therapy compared with conventional therapy in patients with ulcerative colitis. (n/N = patients with response/total patients).

Figure 5.

 Sensitivity analysis for response rates in heparin supplementation therapy compared with conventional therapy (n/N = patients with response/total patients).

Figure 6.

 Sensitivity analysis for response rates in heparin supplementation therapy compared with conventional therapy in patients with ulcerative colitis (n/N = patients with response/total patients).

Publication bias

Funnel plots were also provided to assess the publication bias. To systematically evaluate the publication bias of this meta-analysis, we provided funnel plot of all pooled studies (Figure 7). We will find the funnel plot was slight asymmetrical distribution. In the study by Panes et al.,11 heparin was used as monotherapy without concomitant conventional treatment, and we cannot find any studies indicating significant difference between heparin and conventional treatment group. All of the above suggested some patterns for publication bias for the main summary estimate of this meta-analysis.

Figure 7.

 Funnel plot analysis comparing heparin therapy with conventional therapy for response rates.

Discussion

It was suggested in our meta-analysis that adjuvant administration of heparin in patients with UC was safe, but no additive benefit over conventional therapy for UC was noted. Neither UH nor LWMH showed additive benefit over conventional therapy for UC even as supplemental therapy. And heparin was no better than placebo for response rate of patients with UC.

The pathogenesis of UC remains unclear despite intense research efforts. A substantial number of patients with acute flares of UC do not respond to conventional therapy, with about 15% of patients undergoing colectomy as a consequence of failed medical therapy.19 Much effort has been taken to find alternative approaches to improve the medical efficacy of UC.20, 21

Heparin has been shown to interfere with recruitment, adhesion and migration of leucocytes and stimulate basic fibroblast growth factor and insulin-like growth factors in vitro and in vivo,22–27 which suggest that heparin, with its anticoagulant and anti-inflammatory properties, can be a kind of potential drug in UC.

Clinical trials have evaluated the effects of both UH and LMWH in ameliorating the symptoms of UC. Low-molecular-weight heparin has a more predictable anticoagulant effect than UH, is easier to administer, and does not require monitoring. LMWH has become the subject of therapy because of lower risk of adverse events and ease of administration recently, which makes this type of heparin a substitute for UH. Our findings indicated that there were no significant differences comparing either UH or LMWH supplementation to conventional therapy in patients with UC.

In some uncontrolled and open-labeled trials, it was noted that patients with UC treated with UH achieved clinical remission.2–4 UH has shown beneficial effects as an adjuvant treatment of UC in open trials, particularly in steroid refractory UC. LMWH offered advantages in the method of administration and showed almost the same benefits.8–10, 28 We only included RCTs to provide more objective results based on the rationale of evidence based medicine.

There are some limitations to our analysis. Firstly, studies in the past couple years have shown that response or remission rates in UC studies can vary widely depending on some of these definitions. In study by Ang et al.,12 remission or CR to therapy was defined by Truelove and Witt Scoring Index of <5, normal stool frequency and consistency, an endoscopic score = 0 or a histopathological grading = 0. A PR was defined by a ≥ 50% reduction in clinical disease activity score and stool frequency and a reduction in endoscopic score or histopathological grading by at least 1. In the study by Bloom et al.,13 patients achieving a colitis activity score of 0 at the end of treatment were defined as complete responders. A PR was defined as a fall in the colitis activity score of at least three points. In the study by Zezos et al.,15 remission or CR to therapy was defined by the coexistence of clinical and endoscopic remission (SCCAI score of ≤ 2 and endoscopic score of ≤ 3). A PR to therapy was defined by a ≥ 50% reduction of SCCAI score and a reduction in endoscopic score by at least one grade. To date, a universally accepted index of remission in UC is still lacking. All of the above indicated the potential heterogeneity in our meta-analysis. Secondly, different study durations would influence their results. In the study by de Bievre et al.,18 at 4 weeks, the mean Clinical Activity Index was 7 (95% CI: 5–9) in both the heparin and placebo group, and at 8 weeks the mean Clinical Activity Index was 5 (95% CI: 3–7) in the heparin group and 6 (95% CI: 3–8) in the patients treated with placebo. Thirdly, the study by Panes et al.,11 showed no response in the heparin group and a significantly higher rate of rectal bleeding. The results have been attributed to some factors including the lack of concomitant conventional therapy and very short treatment period of 10 days. Therefore, we provided three sensitive analyses not only to evaluate the stability, but also to show some aspects of heterogeneity of the meta-analysis. Finally, though Tinzaparin, Deligoparin, Enoxaparin, Reviparin and Dalteparin are all LMWHs in pooled studies, these compounds differ significantly in their biologic properties, which may lead to different results in the treatment of UC. Various LMWHs exhibit specific molecular and structural attributes that are determined by the type of manufacturing process used. Enoxaparin, which is prepared by benzylation followed by alkaline hydrolysis of UH, exhibits a double bond at the non-reducing end and the presence of a unique bicyclic structure namely 1,6 anhydromanno glucose or mannose, or both, at the reducing end. Similarly, the other LMWHs, such as dalteparin, deligoparin, tinzaparin, and reviparin, exhibit specific structural characteristics that may contribute to their own unique biochemical and pharmacological profiles.29

In addition, there was no additive benefit to adjuvant administration of heparin in patients with UC. Though administration of heparin in patients with UC is safe, heparin was no better than placebo for response rate of patients with UC.

Acknowledgements

Declaration of personal and funding interests: None.

Ancillary