An Evidence-Based Review on Medical Therapies for IBD PDF

S2 VOLUME 106 SUPPLEMENT 1 APRIL 2011 nature publishing group An Evidence-Based Systematic Review on Medical Therapies for Inflammatory Bowel Disease Nicholas J. Talley, MD, PhD,...

S2 VOLUME 106 SUPPLEMENT 1 APRIL 2011 An Evidence-Based Systematic Review on Medical Therapies for Inflammatory Bowel Disease Nicholas J. Talley, MD, PhD, FACG1, Maria T. Abreu, Marla C. Dubinsky, MD5, Stephen B. Hanauer, MD, FACG6, Thomas A. Ullman, MD, FACG9 and Paul Moayyedi, MB, ChB, IBD Task Force SUPPLEMENTARY MATERIAL is linked to the online version Am J Gastroenterol 2011; 106:S2–S25; doi:10.1038/ajg.2011.5 Crohn’s disease (CD) and ulcerative colitis (UC) are chronic inflammatory disorders of the gastrointestinal tract. Collec- tively they are termed inflammatory bowel disease (IBD) and it is estimated that 1.5 million Americans (1) suffer from UC and CD. Their etiologies are unknown, although both are thought to arise from a disordered immune response to the gut in genetically predisposed individuals (1). The characteristics of the inflammatory response are different, with CD typically caus- ing transmural inflammation and occasionally associated with granulomas, whereas in UC the inflammation is usually confined to the mucosa. Both UC and CD exhibit a relapsing and remit- ting course and there is a significant, often dramatic, reduction in quality of life during exacerbations of the disease (2). This has an impact on psychological health, with active IBD patients expe- riencing greater levels of distress and feelings of lack of sense of self-control compared with the normal population and patients with inactive IBD (3,4). Extrapolation from US administrative claims databases sug- gests that IBD is responsible for 2.3 million physician visits (5), 180,000 hospital admissions (6), and costs $6.3 billion (7) annu- ally. There have been recent guidelines on the management of both UC (8) and CD (9) that direct the clinician on diagnosis and treatment. Approximately 33% of the cost of IBD is due to medical therapy (7), and given the substantial clinical burden and economic cost of IBD it is important to establish the effectiveness of current medical therapies in both UC and CD. Although there have been several systematic reviews on the efficacy of therapy (10,11), this is a rapidly changing field and there is a need for a comprehensive review of the literature. The American College of Gastroenterology IBD Task Force developed a protocol for systematically reviewing the data on currently available therapies 1 Department of Medicine, Faculty of Health, University Miami, Miller School of Medicine, Miami, Florida, USA; 4 Department of Medicine, University of Manitoba, Winnipeg, Center, Los Angeles, California, USA; 6Section of Gastroenterology 7 Department of Medicine, Mayo Clinic College of Medicine, of California San Diego, La Jolla, California, USA; New York, New York, USA; 10Division of Gastroenterology, Correspondence: Nicholas Talley, MD, PhD, FACG, Department E-mail: [email protected] VOLUME 106 | SUPPLEMENT 1 | APRIL 2011 www.amjgastro.com VOLUME Evidence-Based 106 Review Systematic SUPPLEMENT 1 APRILfor on Medical Therapies 2011 IBD S3 Table 1. Continued Country CD incidence/105 (ref) UC incidence/105 (ref) Japan 1.2 (43) Singapore 3.6 (38) 6.0 (38) India 6.02 (46) Oceania New Zealand 16.5 (25) 7.6 (25) CD, Crohn’s disease; UC, ulcerative colitis. 16.8 (32) also has a high incidence in New Zealand, suggesting that the dis- ease may get more common the further the geographical location 3.9 (34) is from the equator. There may therefore be an environmental 10.0 (36) agent predisposing to CD that is more prevalent in a more tem- perate climate. A similar epidemiological pattern exists for UC (Figure 1b), although the North–South gradient is not as pro- 7.0 (40) nounced (Figure 2b). CD has been increasing in incidence over time in most coun- 9.1 (35) tries (Figure 3a), although it seems to have reached a plateau in the United States (14), Sweden (30), and the UK (48). Incidence rates are also increasing over time for UC, although the degree of change is less significant (Figure 3b). Again, rates appear to have reached a steady state in the United States (14). This could be due to wider availability of diagnostic tests such as colonoscopy, but is more likely to represent a change in the environment that predis- 5.9 (28) poses subjects to develop CD. The nature of this environmental 11.01 (29) agent or agents remains unclear. The peak age of onset for CD has consistently been reported at around 15–30 years (49), although the disease can be diagnosed at any age. The first diagnosis of UC also occurs most commonly at 15–30 years of age and then the incidence plateaus. Some Western studies suggest that CD is more common in women, with a ratio of around 3:2 (48), and a sex hormone influence on risk of CD is supported by a meta-analysis showing an association between oral contraceptive pill use and CD in women (50). However, the increased risk in women is not a consistent finding in the West (30), and in the East CD may be more common in men (43). The incidence of UC does not have a consistent sex difference (20). Caucasians are reported to have a higher risk of IBD than other 4.1 (16) races, but a systematic review suggests that the incidence in Afri- can Americans is approaching that of Whites, and that in Asian Americans and Hispanics the incidence rate is likely higher than was previously thought (51). Other epidemiologic risk factors include cigarette smoking and appendectomy. Systematic review 2.8 (38) data (52) confirm that smoking is a risk factor for CD (9 studies, odds ratio (OR) = 1.76; 95% confidence interval (CI): 1.40–2.22) but is protective for UC (13 studies, OR = 0.58; 95% CI: 0.45–0.75). Passive smoking data are less clear, with one study suggesting no impact on either CD or UC (53) and another study reporting an increased risk of CD with passive smoking in children under 3.08 (42) the age of 12 (54). A meta-analysis (55) reported that appendec- tomy was significantly associated with risk of CD, but there was The American Journal of GASTROENTEROLOGY bowel disease. (a) Incidence rates for Crohn’s disease. (b) Incidence rates for ulcera- dark green 2–3.9, yellow 4–5.9, light orange 6–7.9, dark orange 8–9.9, pink 10–12, red > 12 per consistent protective effect (OR = 0.31; 95% CI: 0.25–0.38), with no significant heterogeneity between results. The rapidly increasing rates of CD and UC seen over the past 20 years point to environmental factors being important in the cause of IBD. Nevertheless, most diseases are due to the interplay between genes and the environment and IBD is no exception. Identical twin studies have shown a concordance rate of 50–60% (57,58). Since the description of the association between NOD2 polymorphisms and CD (59), there has been a plethora of genes VOLUME 106 | SUPPLEMENT 1 | APRIL 2011 www.amjgastro.com VOLUME Evidence-Based 106 Review Systematic SUPPLEMENT 1 APRILfor on Medical Therapies 2011 IBD S5 a Saudi Arabia Puerto Rico Incidence (per 100,000) 10 Iceland Bosnia & Herzegovina Greece Denmark 5 Hungary China Brazil 0 South Korea 1980 1990 2000 Japan Year 60 b Incidence (per 100,000) 20 Puerto Rico Iceland 15 Denmark Hungary 10 Brazil South Korea 5 0 1980 1990 2000 Year Figure 3. Changes in inflammatory bowel disease incidence rates over time. (a) Incidence of Crohn’s disease over time. (b) Incidence of ulcera- tive colitis over time. 40 60 PC2 rates with de- disease incidence colitis incidence Ulcerative colitis PC1 Healthy Crohn’s disease P value: 0.031 Figure 4. Gut microbiome in Crohn’s disease, ulcerative colitis, and healthy subjects (reproduced with permission from ref. (63)). for Salmonella and Campylobacter infections. There is a risk of case ascertainment bias in these studies, as subjects who are known to have these infections are likely to get more follow-up investigations than those who do not have these infections. The Metagenomics of the Human Intestinal Tract Consortium has obtained fecal samples from 124 Danish and Spanish subjects and sequenced their gut flora (63). Microbial genes in these samples were 150-fold greater than the human genome and over 99% were bacterial. This group found that UC, CD, and healthy controls all had distinct bacterial patterns (63) (Figure 4). CD patients had a less diverse microbial flora and although the number of patients evaluated was small, this The American Journal of GASTROENTEROLOGY Section 2 Methodology of the ACG Task Force on medical therapies for IBD There are a variety of medical therapies available to treat both UC and CD, including 5-aminosalicylic acid (5-ASA) drugs, cortico- steroids, immunosuppressive agents, biological therapies, and anti- biotics. These agents can heal active disease, prevent relapse, and improve the quality of life (78,79), but the relative efficacy of these competing therapies is unclear. There have been a number of sys- tematic reviews (10,11,80–96) that have evaluated randomized con- trolled trials (RCTs) of these interventions, but some need updating and some used end points that are poorly defined (e.g., “response to therapy”). Furthermore, we have shown that errors were made in identifying RCTs, assessing eligibility, and extracting data in all previous systematic reviews of irritable bowel syndrome (IBS) (97) when evaluating the evidence for an updated ACG monograph on medical therapies for IBS (98). It is possible that similar errors have been made in some systematic reviews of IBD therapies. We therefore completed a series of systematic reviews evaluating the efficacy of medical therapies for CD and UC in inducing remis- sion and preventing relapse using a similar methodology to the previous ACG monograph on IBS management (98). Systematic review methodology Full details of the methodology and results of the systematic reviews are given in individual papers that accompany this mono- graph (99–104). Only parallel-group RCTs or the first phase of crossover randomized trials were eligible for inclusion in the review. Studies needed to have evaluated adult patients with either active or quiescent CD or UC. The following interventions were considered: important in (a) Oral 5-ASA drugs (e.g., sulfasalazine, mesalamine) (b) Oral traditional corticosteroids (e.g., prednisone or pred- nisolone) (c) Oral budesonide (d) Immunosuppressive therapy (e.g., azathioprine, 6-mercap- topurine (6-MP), methotrexate) (e) Biological therapies (e.g., infliximab, adalimumab, etanercept, certolizumab) (f) Antibiotic therapy (e.g., antimycobacterial drugs, metronidazole) consistent These interventions were compared with placebo or with no intervention. The minimum duration of therapy was 14 days for treatment of active disease and 6 months for prevention of relapse, and any dose of drug was considered eligible. Trials evaluating medical therapy in active CD were included if they reported evidence of clinical remission or healing of fistula as an outcome. Remission was defined according to a prospectively agreed hierarchy, and in the event that more than one outcome was reported the outcome that was most stringent according to our protocol was selected. The hierarchy agreed among the group was as follows: (a) Crohn’s Disease Activity Index (CDAI) < 150 (or other validated index) VOLUME 106 | SUPPLEMENT 1 | APRIL 2011 www.amjgastro.com VOLUME Evidence-Based 106 Review Systematic SUPPLEMENT 1 APRILfor on Medical Therapies 2011 IBD S7 various reductions in the Crohn’s Disease Activity Index or “reduc- tion in fistula drainage” for CD (105) or Disease Activity Index for UC (106). Furthermore, we realize that the duration of placebo-control- led trials for induction and maintenance of IBD may be insufficient to evaluate longer-term risks of serious infections, immunological events, and malignancies. Thus, we attempted to interpret and ana- lyze the more stringent, systematic remission and adverse-event data within the context of a broader clinical perspective. Relevant RCTs were identified using MEDLINE (1966 to Decem- ber 2010), EMBASE (1984 to December 2010), Cochrane Central Register of Controlled Trials (Issue 4, 2010), and the Cochrane Inflammatory Bowel Disease Group Specialized Trials Register. No language restrictions were used. We conducted a recursive search of the bibliography of all relevant trials and papers identified by the electronic search strategy and evaluated abstracts from DDW 2002–2010 and UEGW 2002–2010. Each potentially eligible paper was evaluated by two independ- ent investigators according to pre-stated eligibility criteria, and where disagreements occurred the opinion of a third investigator was obtained. Papers that were not eligible are outlined in the Sup- plementary Appendix online. Each eligible study was evaluated for seven components of risk of bias as described in the Cochrane handbook (107), including method of randomization, concealment of allocation, and implementation of masking. Data were extracted on specially developed electronic forms by two independent reviewers and any discrepancy was resolved by consensus. Analyses were conducted separately for CD, UC, treatment of active disease, and prevention of relapse. Each drug class was com- pared with placebo. The last time point of assessment was used for analysis. For trials that evaluated treatment of active disease, benefit was expressed in terms of the relative risk (RR) of remaining with active disease (with 95% CIs) as the summary outcome statistic, while trials that assessed prevention or relapse were expressed in terms of the RR of relapse. Intention-to-treat data were used, with any drop-outs/withdrawals from the randomized groups being assumed to be treatment failures. Data were synthesized using a fixed-effects model, but if significant heterogeneity was found (I2 > 20% and/or chi-squared P > 0.15) a random-effects model (108) was used. Adverse effects were summarized with RRs, comparing the incidence rates of the (as well as overall proportion with) adverse effect(s). Number needed to treat (NNT) and number needed to harm (NNH) were calculated using the following formula: 10 NNT = RRR × BR (RRR = relative risk reduction, BR = baseline risk). This was checked by inversing the pooled risk difference from the meta-analysis. Funnel plots were produced for the principal outcome for each comparison, and Egger’s test (109) of funnel plot asymmetry was used, to investigate whether publication or other bias may have adversely affected the results. Heterogeneity was assessed using χ2 statistic and I2. A value of I2 > 20% was considered to indicate heterogeneity and the reasons were explored using subgroup analyses. The American Journal of GASTROENTEROLOGY as SASP at healing distal UC, but disease remained active with sulfapyridine (113). The beneficial effects of SASP on UC may therefore be due to the mesalamine moiety and this has led to other 5-ASA formulations and delivery systems (114,115). The precise mechanism of action of these drugs is unclear, but they are believed to act as anti-inflammatory agents, as they inhibit nuclear factor kappa B and chemoattractant leukotrienes and alter pros- taglandin metabolism (116). The systematic availability of 5-ASA for all preparations is low (117), so adverse events are generally no greater than placebo in RCTs (99,100,118). However, there are very rare serious adverse effects of 5-ASA, such as interstitial nephritis, pancreatitis, pneumonitis, pericarditis, and hepatitis (119). the bene- Section 3.1 Efficacy of 5-ASA therapies at inducing remission in active UC 5-ASA therapies are effective at inducing remission in mild-to- moderately-active UC Recommendation: strong. Quality of evidence: moderate. A weak The systematic review is described in detail elsewhere (99). There were 11 RCTs evaluating 2,086 patients (120–130) that compared 5-ASA therapies vs. placebo. The majority of studies evaluated mesalamine and enrolled patients with mild-to-moderately-active UC. There was a strong effect in favor of 5-ASA therapy, with a NNT of 6 (95% CI: 5–8) (99), and with 40% of patients achieving remission overall in the active treatment group compared with 20% in the placebo group. The quality of the studies was graded as mod- erate, as there was some heterogeneity in the data, some funnel plot asymmetry suggesting publication bias or other small study effects and only two studies (121,123) having a low risk of bias. This would normally have made the quality of evidence low, but the grading was increased to moderate as the effect of 5-ASA on active UC was strong and the heterogeneity was largely explained by duration of therapy and definition of treatment success, with studies using the most stringent definition of success (mucosal healing) showing least heterogeneity. There was no difference in efficacy between the different 5-ASA preparations using either direct (data not shown) or indirect comparisons. There were two trials (120,121) involv- ing 84 patients evaluating SASP with an NNT of 3 (95% CI: 2–10) and seven trials (123–128,130) involving 1,722 patients evaluating mesalamine with an NNT of 6 (95% CI: 5–9). The optimum dose for 5-ASA therapies appears to be 2.4 g per day of mesalamine or equivalent, with no apparent benefit from increasing the dose. There were 10 trials involving 2,414 patients (124,126–128,130–135) that compared high-dose (defined as more than 2.5 g per day) with standard-dose (defined as 2–2.5 g per day) 5-ASA in treating active UC (99). The majority of patients had mild-to-moderate UC with treatment durations of 4–8 weeks. There was no statistically significant difference between the two groups, with 31% achieving remission in the high-dose compared with 30% in the standard-dose group. Clinicians may still wish to increase the dose of 5-ASA beyond 2.4 g per day in patients on maintenance therapy with a mild relapse of UC, but it is possible that any response obtained is due to the placebo effect or regres- sion to the mean. VOLUME 106 | SUPPLEMENT 1 | APRIL 2011 www.amjgastro.com VOLUME Evidence-Based 106 Review Systematic SUPPLEMENT 1 APRILfor on Medical Therapies 2011 IBD S9 planned secondary analysis was remission or improvement in CD activity, and in this analysis mesalamine demonstrated a statisti- cally significant effect in improving CD (NNT = 7; 95% CI: 4–20). Overall we recommended against using 5-ASA to induce remission in active CD. The reason for this was that the data were equivocal and this did not include the Crohn’s III trial (160) involving 310 patients. Data on CD remission or improvement were not avail- able for this trial, but the mean CDAI scores were similar between 5-ASA and placebo arms (160), suggesting that this trial is likely to be negative. Given that this trial represents 33% of all the patients in the meta-analysis, we felt that it was likely that 5-ASA was not effective at inducing remission in active CD. 5-ASA therapies are likely to be maximally effective in the colon and yet the majority of patients in the studies analyzed had either ileal or ileocolonic disease. The quality of evidence was categorized as low as there was only one trial (159) with low risk of bias, results were heterogene- ous, and conclusions were different depending on what outcome measure was chosen. Section 3.4 Efficacy of 5-ASA therapies at preventing relapse in quiescent CD 5-ASA therapies are not recommended for preventing relapse in quiescent CD Recommendation: weak. Quality of evidence: low. There were 16 RCTs (155,159,161–174) assessing 2,496 patients comparing 5-ASA with placebo or no therapy in quiescent CD for 6 to 48 months. The majority of patients had ileal or ileoco- lonic disease and there was no significant difference in relapse rates between the two groups. The overall relapse rate was 56% with 5-ASA therapies and 57% in the control group (100). In the 11 trials that evaluated mesalamine, there was a trend toward benefit of 5-ASA, but this was not statistically significant (100). There were only three trials (155,165,170) that had low risk of bias and a subgroup analysis of these studies suggested a small but statistically significant benefit of 5-ASA at preventing CD relapse (NNT = 13; 95% CI: 6–100). The evidence was of low quality as there was a paucity of trials that had a low risk of bias, there was unexplained heterogeneity between studies and con- clusions differed in those with unclear vs. low risk of bias. Overall the evidence available was not sufficient to recommend 5-ASA therapies to prevent CD relapse. Unlike the recommendation of the Cochrane review (91), we concluded that further trials might be helpful in determining whether 5-ASA therapies have a role in colonic CD. Section 4 Efficacy of corticosteroid therapies in IBD Corticosteroids are another example of drugs that were first developed for rheumatoid arthritis that then were applied to IBD. Philip Hench and Edward Kendall from the Mayo clinic announced the development of “compound E” (175) that, when injected into a rheumatoid arthritis joint, allowed the patient a miraculous recovery. This wonder drug is today known as cor- tisone and the medical establishment greeted this discovery so The American Journal of GASTROENTEROLOGY Section 4.2 Efficacy of corticosteroid therapies at inducing remission in active CD Standard corticosteroids therapies are effective at inducing remission in active CD Recommendation: strong. Quality of evidence: low. drugs had We identified two RCTs (155,159) involving 267 patients that com- pared oral corticosteroids with placebo in active CD. Each trial reported a statistically significant effect in favor of corticosteroids, with results varying between a NNT = 2 (95% CI: 1.6–3) (159) and a NNT = 5 (95% CI: 3–14) (155). Overall 60% achieved remission with corticosteroid therapy compared with 31% in the placebo group. As the estimate of effect was so different and the number of patients studied was modest, this meant that when the two tri- als were combined the effect of corticosteroids was not statistically significantly different from placebo in a random effects model using RR. When risk difference was used as the summary statistic in the meta-analysis, there was a significant effect of corticosteroid (NNT = 3; 95% CI: 2–11). There were no trials evaluating intrave- nous therapy and data available were not sufficient to evaluate the efficacy according to severity of disease. The evidence quality was very low as there was a paucity of evidence, heterogeneity between the two studies and neither study was at low risk of bias. We did, however, make a strong recommendation that standard corticos- teroids were effective in active CD as there is evidence that budeso- nide is more effective than placebo at inducing remission in CD and there is also evidence from RCTs that systemic corticosteroids are more effective than budesonide (see below). Budesonide therapy is effective at inducing remission in mild-to- moderately-active CD remission in active UC Recommendation: strong. Quality of evidence: low. There were two RCTs (195,196) involving 458 patients that com- pared budesonide with placebo in active CD. Patients had termi- nal ileal and/or right-sided colonic disease. Budesonide was more effective than placebo at inducing remission with a NNT = 5 (95% CI: 3–9) (101). Overall 45% of CD patients achieved remission with budesonide compared with 24% on placebo. Both trials gave similar results with no heterogeneity between the two studies. However, the evidence quality was low as neither trial had a low risk of bias and the number of patients studied was modest. thought to act Standard corticosteroids are more effective than budesonide at inducing remission in mild-to-moderately-active CD Recommendation: weak. Quality of evidence: low in the The systematic review (101) identified six trials (197–202) evalu- ating 669 patients that compared standard corticosteroids vs. budesonide in active CD. The majority of patients had terminal ileal or right-sided colonic disease. Trials evaluated treatment after 8–10 weeks, with 62% achieving remission in the standard corti- costeroid group vs. 53% in the budesonide group. The NNT for standard corticosteroids over budesonide at inducing remission was 11 (95% CI: 6–50). Data on overall adverse events were only VOLUME 106 | SUPPLEMENT 1 | APRIL 2011 www.amjgastro.com VOLUME Evidence-Based 106 Review Systematic SUPPLEMENT 1 APRILfor on Medical Therapies 2011 IBD S11 These drugs had already been used in inflammatory conditions once the medical community realized that the dramatic effects of corticosteroids were often transient or required long-term ther- apy with serious adverse consequences. Again, rheumatologists led the way with descriptions of the benefits of methotrexate on rheumatoid arthritis in the early 1950s (210). Gastroenterologists followed suit with case reports of azathioprine and 6-MP improv- ing IBD (211–214). Although the mechanism of action differs between these three classes of medications, collectively they are referred to as immunosuppressants due to their direct or indirect effects on immune cell number or function. The specific role of immunosuppressants in IBD as induction agents and as main- tenance therapy needs to be re-addressed, as systematic reviews assessing their efficacy have been limited by use of ambiguous end points (clinical response), vague search and inclusion criteria, or other potentially problematic methodologies. We therefore con- ducted an updated systematic review (102) of RCTs of azathio- prine, 6-MP, methotrexate, tacrolimus, and cyclosporin compared with placebo with clear end points for induction of remission or maintenance of remission in UC and CD. Owing to the differ- ences in their mechanism of action, we elected to perform sepa- rate analyses for thiopurines, methotrexate, and the calcineurin inhibitors (102). Adverse effects are poorly described in randomized trials, which are also not the best design for determining rare but serious long- term complications. Observational data provide more information on adverse effects. Azathioprine and 6-MP are associated with nausea, allergic reactions, acute pancreatitis, hepatitis, increased risk of infection, malignancy, and bone marrow suppression (215). As bone marrow suppression can occur at any point during ther- apy, it is recommended that patients have regular blood counts to monitor for leucopenia. The adverse effects of methotrexate include hepatotoxicity, pneumonitis, increased risk of infection, malignancy, alopecia, stomatitis, and myelosuppression. Hepato- toxicity risk increases when the accumulated dose of methotrex- ate exceeds 1.5 g (215), but seems to be minimal at doses less than 5 g (216). The main adverse effect of cyclosporin and tacrolimus is renal toxicity and drug levels should be monitored (215). Other side effects include hypertension, hirsuitism, headache, opportun- istic infections, seizures, and paresthesia (215). corticos- Section 5.1 Efficacy of immunosuppressant therapies at inducing remission in active UC Azathioprine and 6-MP are not recommended for inducing remis- sion in active UC Recommendation: weak. Quality of evidence: very low. A systematic review of the literature (102) for this monograph revealed only two RCTs (217,218) evaluating 130 patients. Both RCTs (217,218) were negative, and although there was a trend to benefit in the meta-analysis this was not statistically significant. There was no heterogeneity between studies, but neither RCT had a low risk of bias. We felt azathioprine and 6-MP should not be generally used to induce remission in UC, but the recommenda- tion was weak as it was recognized that only a small number of The American Journal of GASTROENTEROLOGY the azathioprine group compared with placebo, with an NNT of 4 (95% CI: 2–10). The annual relapse rate was 39% in the azathio- prine group compared with 66% in the placebo group. There was little heterogeneity between studies, but the evidence was graded as low as there were no low-risk-of-bias trials and the number of patients studied was small and so the CIs were wide. There was one additional RCT (226) that evaluated azathioprine withdrawal in 79 patients who had been maintained on this drug. This study found that relapse occurred in 36% of the patients who continued on azathioprine and in 59% of the patients in the placebo group after 1 year and concluded that thiopurines should be continued for at least 2 years if they appear to be effective. corticosteroids. Methotrexate is not recommended for preventing relapse in quiescent UC. Recommendation: weak. Quality of evidence: low. There were two RCTs (219,227) with a total of only 58 patients that compared methotrexate with placebo in quiescent UC. Both used oral methotrexate at a dose of 12.5 (219) or 15 mg (227) weekly and followed patients up for 9–12 months. Both trials were negative. The meta-analysis (102) demonstrated no statistically significant effect of methotrexate and no trend toward benefit. It is possible “response” was that subcutaneous or intramuscular methotrexate and/or a higher dose of 25 mg weekly may be more beneficial in preventing relapse in UC, but with the current evidence the effectiveness of this approach is unproven. patient group to Section 5.3 Efficacy of immunosuppressant therapies at inducing remission in active CD Azathioprine and 6-MP are not recommended for inducing remis- sion in active CD. Recommendation: weak. Quality of evidence: low. group, The systematic review identified five RCTs (158,219,228–230) eval- uating 380 patients and comparing azathioprine or 6-MP with pla- cebo in active CD. Thiopurines were not statistically significantly better than placebo at inducing remission. One trial evaluated 6- MP (219) and none were at low risk of bias. There was only minor heterogeneity between studies, and all but one (228) was negative. Two RCTs (231,232) were excluded as they only reported symptom improvement rather than CD remission, but including these studies did not alter the result and both studies reported a lack of efficacy of azathioprine. One study (233) evaluating 6-MP in 83 patients was excluded, as it only reported on symptom improvement after 1 year. This trial was very positive in favor of 6-MP, but the relevance of this is unclear as data were only provided at 12 months. Intramuscular methotrexate is effective at inducing remission in active CD. Recommendation: weak. Quality of evidence: low. 25) There were only two RCTs (234,235) involving 193 patients that compared methotrexate with placebo in active CD. Overall there was a strong trend in favor of methotrexate, but this was not VOLUME 106 | SUPPLEMENT 1 | APRIL 2011 www.amjgastro.com VOLUME Evidence-Based 106 Review Systematic SUPPLEMENT 1 APRILfor on Medical Therapies 2011 IBD S13 two RCTs (245,246) of post-operative CD patients that suggested that azathioprine (245) or 6-MP (246) was superior to placebo at preventing post-operative recurrence. Methotrexate is effective at preventing relapse in quiescent CD Recommendation: weak. Quality of evidence: low. There was one trial (247) that compared methotrexate with pla- cebo in quiescent CD. This trial enrolled 76 patients and had an unclear risk of bias. Relapse rates were significantly lower in the methotrexate group (NNT = 4; 95% CI: 2–25). Another trial (248) evaluated methotrexate vs. placebo in 33 corticosteroid-dependent CD patients and found a trend toward lower relapse rates at 1 year as corticosteroids were withdrawn in the methotrexate group. Cyclosporin is not recommended for preventing relapse in quiescent CD Recommendation: weak. Quality of evidence: low. We identified only one RCT (237) that evaluated cyclosporin vs. placebo to prevent CD recurrence in 118 patients. There was no statistically significant benefit of the active drug, with over 70% relapsing at 1 year in both groups. to rec- Section 6 Efficacy of biological therapies in IBD Anti-tumor necrosis factor alpha antibodies (anti-TNFα) were originally developed to improve survival from septicemia, but a phase II trial was negative (249). Feldman and Maini at the Kennedy Institute of Rheumatology in London highlighted the importance of cytokines in rheumatoid arthritis and managed to convince the pharmaceutical industry to give them some anti- TNFα antibodies for a small open-label clinical trial. Their results were impressive (250) and, as with previous rheumatological dis- coveries, they were soon used in IBD. Biological therapies were introduced into the United States, and subsequently the world market, for the treatment of CD in 1998 and eventually for the treatment of UC. These therapies have been incorporated into the recent guidelines for therapy of CD by the American College of Gastroenterology (9), American Gastroenterological Association (251), and The European Crohn’s and Colitis Organization (252), and for UC by the American College of Gastroenterology (8). Such guidelines have been formulated on the basis of composite evidence from clinical trials, clinical series, and expert opinions. Previous meta-analyses have examined the benefit of these bio- logical therapies in various situations, but none have synthesized all current available evidence for their role in IBD, and some have important limitations. We have therefore conducted a systematic review and meta-analysis of RCTs to estimate the efficacy and safety of these drugs in IBD as a whole (103) and focused on dou- ble-blind, placebo-controlled trials to minimize bias and hetero- geneity between trials. Overall there were no statistically significant differences between adverse events in patients randomized to biological ther- apy compared with placebo. However, as mentioned previously, The American Journal of GASTROENTEROLOGY Section 6.3 Efficacy of biological therapies at inducing remission in active CD Anti-TNF antibody therapies (infliximab, adalimumab, and cer- tolizumab pegol) are effective at inducing remission in ambulatory patients with moderate-to-severely-active CD Recommendation: strong. Quality of evidence: moderate. concerns The systematic review (103) identified 10 RCTs (259–268) that evaluated 2,756 patients and compared anti-TNFα antibody ther- apy with placebo in active CD. Remission of CD was achieved in 28% of patients randomized to receive anti-TNFα antibodies at 4–12 weeks, compared with 19% of patients randomized to pla- cebo. The NNT was 8 (95% CI: 6–17) (103). The impact of anti- TNFα antibody therapy on active CD appears modest compared with some other therapies described above, but from a clinical perspective the data are impressive considering the “refractori- ness” of patients enrolled and the short duration used to assess the response. Clearly, more patients benefited based on the crite- ria used to assess clinical responses in these trials than achieved clinical remission. Further, based on maintenance of responder trials (see below), many patients may require longer duration of anti-TNF therapy to achieve remissions based on CDAI and, in addition, corticosteroid-free remissions were not assessed in these short-term trials. Based on all these factors, we gave a strong recommendation. The quality of evidence was moderate as there was heterogeneity in the data. This was unexplained but largely driven by one trial (260). Two trials (260,262) had a low risk of bias and both reported a statistically significant effect in favor of biologic therapy, with a similar or smaller NNT than the overall meta-analysis. There was statistically significant heterogeneity between the three different types of anti-TNFα antibodies and this appeared to be driven by more positive results of infliximab and adalimumab compared with certolizumab pegol trials (103). However, such comparisons may be invalid, as trials were performed sequentially rather than concurrently and assessed induction end points at dif- ferent time frames, and, although patients with relatively similar disease activities were enrolled, there were potential biases based on subtle differences in concomitant medications and “refractoriness to corticosteroids or immunosuppressives,” disease duration, and study sites. The recognition of individual formulations response at any particular time point likely reflects the specific time frames and dosing of individual trials and, in the absence of head-to-head comparisons between agents, should not be interpreted as a “faster onset of action” for any anti-TNF biological agent in luminal CD. The evidence from this systematic review and meta-analysis (103) supports the use of biological therapies in patients with luminal CD who have failed treatment with first- and second-line agents, or who are corticosteroid-dependent. No head-to-head comparisons between biological agents have been performed and it is important to consider the data from the large clinical trials that were, primarily, designed to obtain regulatory approval as “minimal” estimates of benefits, as, in clinical practice, dose or dose–frequency adjustments are often required to sustain adequate pharmacokinetics of each agent. We also included multiple doses VOLUME 106 | SUPPLEMENT 1 | APRIL 2011 www.amjgastro.com VOLUME Evidence-Based 106 Review Systematic SUPPLEMENT 1 APRILfor on Medical Therapies 2011 IBD S15 4 (95% CI: 3–5). There were no statistically significant differences in indirect comparisons between inflixmab, adalimumab, and certolizumab. The relapse-preventing potential of the anti-TNFα antibodies is related to the pharmacokinetics of the individual formulation and is influenced by dose, frequency of administration, and immuno- genicity (283). Within the confines of controlled clinical trials there is less potential to adjust dosing or dose frequency to “optimize” maintenance therapies. Hence, it is likely that the estimates of ben- efit reflect minimal effects that can be improved upon in clinical practice (284). In addition, while no apparent difference has been

An Evidence-Based Review on Medical Therapies for IBD PDF

Document Details

Tags

Related

Summary

Full Transcript