Systematic review: infliximab therapy in ulcerative colitis

Authors


Dr J. P. Gisbert, Playa de Mojácar 29, Urb. Bonanza, 28669 Boadilla del Monte, Madrid, Spain.
E-mail: gisbert@meditex.es

Abstract

Summary

Aim

To perform a systematic review and meta-analysis on the efficacy and tolerance of infliximab in ulcerative colitis.

Methods

Selection of studies: evaluating efficacy of infliximab in ulcerative colitis. For the meta-analysis, randomized clinical trials comparing infliximab vs. placebo/steroids. Search strategy: electronic and manual. Study quality: independently assessed by two reviewers. Data synthesis: meta-analysis combining the odds ratios (OR).

Results

Thirty-four studies (896 patients) evaluated infliximab therapy in UC, with heterogeneous results. Mean short-term (2.3 weeks) response and remission with infliximab was 68% (95% CI 65–71%) and 40% (36–44%). Mean long-term (8.9 months) response and remission was 53% (49–56%) and 39% (35–42%). Five randomized double-blind studies compared infliximab with placebo, the meta-analysis showing an advantage (P < 0.001) of infliximab in all endpoints (short-/long-term response/remission): ORs from 2.7 to 4.6, and number-needed-to-treat (NNT) from 3 to 5. Similar infliximab response was calculated independently of the indication (steroid-refractory/non-steroid-refractory) or the dose (5/10 mg/kg). Adverse effects were reported in 83% and 75% of the infliximab and placebo-treated patients (OR = 1.52; 95% CI 1.03–2.24; number-needed-to-harm (NNH) was 14).

Conclusion

Infliximab is more effective than placebo, with an NNT from 3 to 5, for the treatment of moderate-to-severe UC, achieving clinical remission in 40% of the patients at approximately 9 months of follow-up. Further studies are necessary to confirm the long-term efficacy of infliximab in ulcerative colitis.

Introduction

The introduction of steroid therapy in 1955 reduced the mortality of severe ulcerative colitis (UC) to 7%, compared with 24% in the placebo group,1 and it is now <1% in specialist centres.2 Nevertheless, the response of severe UC to steroids has remained unchanged for 50 years,2 and, therefore, severe to moderately severe attacks of UC still have a high colectomy rate.3 On the other hand, a significant proportion of patients will become resistant to steroids.4 Thus, although most patients with UC initially respond to corticosteroids, at 1 year approximately 25% become steroid dependent, and 30% require operation.4

The remaining options in patients with severe UC include cyclosporine or surgery. Although cyclosporine can be effective in a relatively high proportion of these steroid-resistant patients, most of those who do respond initially will eventually need a colectomy.5–7 Furthermore, cyclosporine requires initial continuous parenteral administration and close monitoring to maintain therapeutic blood levels and has well-recognized important side-effects.5–7 Therefore, new therapies with higher efficacy, better safety profile and easier to use are clearly needed for the treatment or refractory UC patients.

Cytokines play a central role in modulating inflammation, and they may, therefore, be a logical target for inflammatory bowel disease therapy using specific cytokine inhibitors.8–11 Tumour necrosis alpha (TNF-α) is a key proinflammatory cytokine in Crohn's disease and in other chronic inflammatory conditions including rheumatoid arthritis and psoriasis.8–10 Infliximab (Remicade®, Centocor) is an intravenously administered chimeric monoclonal immunoglobulin G1 antibody to TNF-α.8–10 Several randomized controlled trials have shown infliximab to be an effective new therapy for Crohn's disease patients with moderate–severe and fistulizing disease, not responding to conventional drug treatment.8–10, 12–15 Thus, infliximab was approved by the US Food and Drug Administration (FDA) in 1998 for the treatment of active, as well as fistulizing Crohn's disease.8–10

Most clinical programmes with infliximab have been focused on Crohn's disease, based on the belief that this condition is a typical Th1-type disease driven by proinflammatory cytokines such as TNF-α.16 Unlike Crohn's disease, UC has long been considered to be a Th2-type of disease with a less prominent role of TNF-α.16 However, although the Th2-type of immune response is predominant in UC, it has been shown that TNF-α may also play a role in its pathogenesis. Thus, there is good theoretical evidence to test infliximab also in UC.

However, only a few studies – most of them including a low number of patients – have evaluated the efficacy of infliximab for the treatment of UC, and the results are, sometimes, controversial. Furthermore, controlled trials (both placebo-controlled or steroids-controlled) have been only exceptionally conducted. Therefore, the primary aim of the present study was to systematically review the efficacy of infliximab for the treatment of UC, and to conduct a meta-analysis of randomized clinical trials comparing the efficacy of infliximab vs. placebo or steroids in patients with UC. Our secondary aim was to perform a meta-analysis of the tolerance of infliximab in controlled studies also in patients with UC.

Methods

Selection of studies

Studies evaluating efficacy of infliximab for the treatment of UC were included in this systematic review. For the meta-analysis, randomized clinical trials comparing the efficacy of infliximab vs. placebo or steroids in patients with UC were included. Only studies with clear information as to the number of patients treated in each therapeutic group, and outcome variable (response or remission rate, see later) clearly stated (separately identified for each therapy), were included.

Search strategy for identification of studies

Trials were identified by searching the Cochrane Library (Issue 4 – 2005), MEDLINE, EMBASE, CINAHL and ISI Web of Knowledge up to January 2006. A search strategy was constructed by using a combination of the following words: (infliximab OR ‘anti-tumor necrosis factor’ OR ‘tumor necrosis factor’ OR ‘tumor necrosis factor alpha antibody’ OR ‘tumor necrosis factor antibody’) AND ‘ulcerative colitis’. Articles published in any language were included. Reference lists from the trials selected by electronic searching were handsearched to identify further relevant trials. We also conducted a manual search of abstracts from 2000 to 2005 from the ‘American Digestive Disease Week’ (DDW), and the ‘United European Gastroenterology Week’ (UEGW) congresses. Abstracts of the articles selected in each of these multiple searches were reviewed and those meeting the inclusion criteria were recorded. References of reviews on inflammatory bowel disease treatment with infliximab, and from the articles selected for the study, were also examined for articles meeting inclusion criteria. In case of duplicate reports, or studies obviously reporting results from the same study population, only the latest published results were used.

Assessment of study quality

The quality of the studies was assessed using the score proposed by Jadad et al.17 based on three items: (i) randomization; (ii) double blinding; and (iii) description of withdrawals and dropouts. The items were presented as questions to elicit yes or no answers. Points awarded for items 1 and 2 depended on the quality of the description of the methods to generate the sequence of randomization and/or on the quality of the description of the method of double blinding. The third item, withdrawals and dropouts, was awarded as zero points for a negative answer and one point for a positive. For a positive answer, the number of withdrawals and dropouts and the reasons had to be stated in each of the comparison groups. Quality assessment of studies was done independently by two reviewers. Discrepancies in the interpretation were resolved by consensus.

Data extraction

The following variables were extracted in a predefined data extraction form (see Table 1): author, year of publication, country where the study was conducted, design (controlled or uncontrolled), number of patients included, study population (children or adults), severity of disease (as defined by the primary studies), indication for infliximab use (corticorefractory or corticodependent UC), treatment with immunosuppressors (including azathioprine, 6-mercaptopurine, methotrexate, cyclosporine or tacrolimus) before and after starting infliximab treatment, dose of infliximab (in mg/kg), number of infliximab infusions (initial infliximab induction regimen and repeated infusions during follow-up), response and remission (see later for definitions) at the short term and at the long term (also see later). In addition, in studies included in the meta-analysis, the following variables were also extracted (see Tables 2 and 3): design (open-label or double-blind), quality score (see Jadad score in previous section, including items of randomization, double blinding and description of withdrawal/dropouts), and short-term and long-term response/remission with infliximab and with placebo or steroids.

Table 1.   Studies evaluating the efficacy of infliximab in patients with ulcerative colitis
ReferenceYearCountryDesignnPopulation Severity (%) Indication (%) IS pre-tx (%)DoseNumber infusions (%)Repeat infusions IS post-tx (%) Short-term response (%)Time (week) Long-term response (%)Time (months)Comments
  1. n, number of patients; CR, corticorefractory; CD, corticodependent; IS pre-tx, immunosuppressive therapy (including azathioprine; 6-mercaptopurine, methotrexate, cyclosporine or tacrolimus) previous to starting infliximab treatment; IS post-tx, immunosuppressive therapy after starting infliximab treatment; dose, in mg/kg body weight; number infusions, initial infliximab induction regimen; repeated infusions, repeated infusions during follow-up, after remission has been achieved; short-term response, response (both partial or complete response) and remission (only complete response) at the first control; long-term response, response (both partial or complete response) and remission (only complete response) at the last control; time (weeks), time at when short-term response/remission was evaluated (in weeks); time (months), time at when long-term response/remission was evaluated (in months).

Actis et al.192002ItalyUncontrolled8AdultsSevere 6/8 (75)
Moderate 2/8 (25)
CR51: 5/8 (63)
2: 2/8 (25)
3: 1/8 (12)
No3/4 (75)Response 4/8 (50)Response: 2/8 (25)7 
Armuzzi et al.20, 212005ItalyControlled10AdultsModerate- SevereCD53YesRemission 10/10 (100)2Remission 7/10 (70)9.8 
Bermejo et al.222004SpainUncontrolled7AdultsSevere 1/7 (14)
Moderate 6/7 (86)
CR 1/7 (14)
CD 6/7 (86)
7/7 (100)51: 5/7 (71)
3: 2/7 (29)
Yes7/7 (100)Remission 6/7 (86)2Remission 6/7 (86)5.5Repeat infusions in 3/7 (43)
Branciforte et al.232002ItalyUncontrolled12AdultsSevere 3/12 (25)
Moderate 7/12 (58)
Mild 2/12 (17)
CR52NoResponse 8/12 (67)6Response 3/12 (25)2.5 
Castro Fernandezet al.242003SpainUncontrolled1AdultsSevereCR0/1 (0)53No1/1 (100)Remission 1/1 (100)2.2Remission 1/1 (100)5 
Chey et al.25, 262001USAUncontrolled16AdultsSevere 16/16 (100)CR51: 10/16 (62)
2: 6/16 (38)
YesResponse 14/16 (88)2Response 4/16 (25)72nd infusion (in 38% of the patients) at 5 months
Cooper et al.272003USAUncontrolled14AdultsSevere 5/14 (36)
Moderate 9/14 (64)
CR 5/14 (36)
CD 9/14 (64)
8/14 (57)51: 7/14 (50)YesRemission 5/14 (36)11 
De Falco et al.282005ItalyUncontrolled20AdultsCR53YesRemission 16/20 (80)2Remission 13/20 (65)6 
Eidelwein et al.292005USAUncontrolled12ChildrenSevere 3/12 (25)
Moderate 3/12 (25)
CD 8/12 (67)
CR 4/12 (33)
8/12 (67)56 (mean)YesResponse 12/12 (100)
Remission 9/12 (75)
2Response 8/12 (67)
CD 6/7 (86)
CR 1/4 (25)
10.4 
Fleisher et al.302001AustraliaUncontrolled17AdultsModerate–severe0/14 (0)53Yes0/14 (0)Remission 14/17 (82)4Remission 13/17 (76)3 
Frenz et al.312002UKUncontrolled1AdultsSevereCR1/1 (100)51Yes1/1 (100)Remission 1/1 (100)0.14Remission 1/1 (100)24 
Gornet et al.322003FranceUncontrolled18AdultsSevere 19/28 (63)
Moderate11/28 (37)
CR 18/30 (60)
CD 5/30 (17)
14/30 (47)51YesResponse 16/18 (89)
Remission 9/18 (50)
1Response 11/18 (61)
Remission 7/18 (39)
10 
Jarnerot et al.332001SwedenUncontrolled11AdultsSevereCR51No8/11 (73)Response 8/11 (73)1Response 6/11 (55)8 
Jarnerot et al.342005SwedenControlled24AdultsSevere 15/24 (63)
Moderate9/24 (38)
CR 24/24 (100)5/24 (21)51NoResponse 17/24 (71)3 
Jølle et al.352004NorwayUncontrolled10AdultsSevere-ModerateCR51YesRemission 3/10 (34)2Remission 9/10 (90)6.5Up to 3 infusions
Kaser et al.362001AustriaUncontrolled6AdultsSevere 6/6 (100)CR 51No Response 6/6 (100)1Response 4/6 (67)5.7 
Kohn et al.37, 382004ItalyUncontrolled13AdultsSevere 13/13 (100)CR6/13 (46)51: 9/13 (69) 2: 4/13 (31)No6/13 (46)Response 10/13 (77)1Remission 9/13 (69)25.6 
Kountouras et al.39, 402005GreeceUncontrolled8AdultsCD53YesResponse 8/8 (100)
Remission 8/8 (100)
2Response 8/8 (100)
Remission 8/8 (100)
9Reinfusions every 8 weeks
Kugathasan et al.412002USAUncontrolled27Adults (56)
Children (44)
Severe 10/27 (37)CR51YesResponse 12/27 (44)
Remission 6/27 (22)
6 
Lim et al.422005USAUncontrolled18AdultsSevere 11/18 (61)
Moderate 6/18 (33)
Mild 1/18 (5.5)
CR 10/18 (55)
CD 3/18 (17)
51YesResponse 14/18 (78)
Remission 8/14 (57)
2Remission 3/18 (17)24 
Ljung et al.432004SwedenUncontrolled22AdultsCR5Yes3 deaths
Mamula et al.44, 452004USAUncontrolled17ChildrenSevere 6/17 (35)
Moderate10/17 (59)
Mild 1/17 (6)
CR12/17 (71)51YesResponse 14/17 (82)
Remission 6/9 (66)
0.3Response 10/16 (63)9.5 
McGinnis et al.462004USAUncontrolled29ChildrenSevere 29/29 (100)CR5, 10Response 18/29 (62)4Response 5/29 (55)12 
Ochsenkuhn et al.472004GermanyControlled6AdultsSevere0/6 (0)53No0/6 (0)Response 5/6 (83)
Remission 3/6 (50)
3Response 5/6 (83)
Remission 3/6 (50)
3 
Oliva-Hemker et al.482002USAUncontrolled5ChildrenModerate–severe3/5 (60)1YesResponse 5/5 (100)
Remission 2/5 (40)
2 
Probert et al.492003UKControlled23AdultsModerate 23/23 (100)CR 23/23 (100)6/23 (26)52Yes6/23 (26)Response 9/23 (39)6 
Ruiz et al.502004SpainUncontrolled4AdultsSevereCR1/4 (25)51: 1/4 (25)
3: 3/4 (75)
No1/4 (25)Response 4/4 (100)
Remission 3/4 (75)
1.1Remission 3/4 (75)4.3 
Russell and Katz512004USAUncontrolled14ChildrenSevere 9/14 (64)
Moderate 5/14 (36)
CR7/14 (50)53YesRemission 8/14 (57)2.6Remission 8/14 (57)6 
Rutgeerts et al. (ACT 1)52–542005MultinationalControlled243AdultsModerate
Severe
CR 74/243 (30)
non-CR 169/243 (70)
125/243 (51)5, 103YesResponse:
Total: 159/243 (65)
5 mg/kg: 84/121 (69)
10 mg/kg: 75/122 (62)
Remission:
Total:
86/243 (35)
5 mg/kg: 47/121 (39)
10 mg/kg: 39/122 (32)
CR-response: 45/62 (73)
non-CR response: 114/181 (63)
8Response:
Total: 125/243 (51)
5 mg/kg: 55/121 (46)
10 mg/kg: 54/122 (44)
Remission:
Total: 84/243 (35)
5 mg/kg: 42/121 (35)
10 mg/kg: 42/122 (34)
13.5Infusion every 8 weeks
Rutgeerts et al. (ACT 2)54, 552005MultinationalControlled241AdultsModerate
Severe
CR 69/241 (29)
non-CR
172/241 (71)
102/241 (42)
5, 103YesResponse:
Total: 161/241 (67)
5 mg/kg: 78/121 (65)
10 mg/kg: 83/120 (69)
CR: 22/35 (63.3)
10 mg/kg: 83/120 (69)
CR: 22/34 (66)
Remission:
Total: 74/241 (35)
5 mg/kg: 41/121 (34)
10 mg/kg: 33/120 (28)
CR-response: 38/59 (64)
non-CR response: 123/182 (68)
8Response:
Total: 129/241 (54)
5 mg/kg: 57/121 (47)
10 mg/kg: 72/120 (60)
Remission:
Total: 74/241 (35)
5 mg/kg: 31/121 (26)
10 mg/kg: 43/120 (36)
7.5Infusion every 8 weeks
Sands et al.562001USAControlled8AdultsSevere 8/8 (100)CR 8/8 (100)1/8 (13)5, 10, 201NoResponse: 4/8 (50)
5 mg/kg: 2/3 (66)
10 mg/kg: 1/3 (33)
20 mg/kg: 1/2 (50)
2Response 4/8 (50)1.7 
Serrano et al.572001USAUncontrolled3Children3/3 (100)51: 1/3 (33)
3: 1/3 (33)
6: 1/3 (33)
Response 3/3 (100)Response 1/3 (33)1.5 
Sriram et al.582004IndiaUncontrolled1AdultsSevere (toxic)CR0/1 (0)51No1/1 (100)Remission 1/1 100%2Remission 1/1 (100)6 
Su et al.592002USAUncontrolled27AdultsSevere 24/27 (89)
Moderate 2/27 (7)
Mild 1/27 (4)
CR 9/27 (33)
CD 10/27 (37)
9/27 (33)51: 14/27 (52)
≥1: 13/27 (48)
Yes11/27 (40)Response: 16/24 (67)
Remission: 10/24 (42)
CR-response: 3/9 (33)
CD-response: 9/10 (90)
0.6Response: 6/10 (60)
Remission: 5/10 (50)
6Reinfusions at the physician's discretion
Table 2.   Controlled studies comparing infliximab vs. placebo
ReferenceDesignQuality scoreShort-term response with infliximab (%)Short-term response with placebo (%)Long-term response with infliximab (%)Long-term response with placebo (%)
  1. Quality score, Jadad scale from 0 to 5 points (see Methods); short-term response, response (both partial or complete response) and remission (only complete response) at the first control; long-term response, response (both partial or complete response) and remission (only complete response) at the last control; CR, corticorefractory.

Järnerot et al.34Controlled
Randomized
Double-blind
5Response: 17/24 (71)Response: 7/21 (33)
Probert et al.49Controlled
Randomized
Double-blind
5Response: 9/23 (39)Response: 6/20 (30)
Rutgeerts et al. (ACT 1)52–54Controlled
Randomized
Double-blind
5Response: 159/243 (65)
⇒ 5 mg: 84/121 (69)
⇒ 10 mg: 75/122 (62)
⇒ CR: 45/62 (73)
⇒ non-CR: 114/181 (63)
Remission: 86/243 (35)
⇒ 5 mg: 47/121 (39)
⇒ 10 mg: 39/122 (32)
Response: 45/121 (37)
Remission: 18/121 (15)
Response: 125/243 (51)
⇒ 5 mg: 55/121 (52)
⇒ 10 mg: 54/122 (51)
Remission: 86/243 (35)
⇒ 5 mg: 41/121 (34)
⇒ 10 mg: 42/122 (37)
Response: 24/121 (20)
Remission: 20/121 (16)
Rutgeerts et al. (ACT 2)54, 55Controlled
Randomized
Double-blind
5Response: 161/241 (67)
⇒ 5 mg: 78/121 (65)
⇒ 10 mg: 83/120 (69)
⇒ CR: 38/59 (64)
⇒ non-CR: 123/182 (68)
Remission: 74/241 (35)
⇒ 5 mg: 41/121 (34)
⇒ 10 mg: 33/120 (28)
Response: 36/123 (30)
Remission: 7/123 (5.7)
Response: 129/241 (54)
⇒ 5 mg: 57/121 (47)
⇒ 10 mg: 72/120 (60)
Remission: 74/241 (35)
⇒ 5 mg: 31/121 (26)
⇒ 10 mg: 43/120 (36)
Response: 32/123 (26)
Remission: 13/123 (11)
Sands et al.56Controlled
Randomized
Double-blind
4Response: 4/8 (50)
⇒ 5 mg/kg: 2/3 (66)
⇒ 10 mg/kg: 1/3 (33),
⇒ 20 mg/kg: 1/2 (50)
Response: 0/3 (0)Response: 4/8 (50)
Table 3.   Controlled studies comparing infliximab vs. steroids
ReferenceDesignQuality scoreShort-term response with infliximab (%)Short-term response with steroids (%)Long-term response with infliximab (%)Long-term response with steroids (%)
  1. Quality score, Jadad scale, from 0 to 5 points (see Methods); short-term response, response (both partial or complete response) and remission (only complete response) at the first control; long-term response, response (both partial or complete response) and remission (only complete response) at the last control.

Armuzzi et al.20, 21Controlled
Randomized
Open-label
2Response 10/10 (100)
Remission 10/10 (100)
Response 10/10 (100)
Remission 10/10 (100)
Response 7/10 (70)
Remission 7/10 (70)
Response 8/10 (80)
Remission 8/10 (80)
Ochsenkuhn et al.47Controlled
Randomized
Open-label
3Response: 5/6 (83)Response: 6/7 (86)Response: 5/6 (83)
Remission: 3/6 (50)
Response: 6/7 (86)
Remission 5/7 (71)

Regarding assessment of tolerance, adverse events in infliximab-controlled studies were extracted (Table 4), including the number of patients that suffered any kind of adverse event, the number of mild-moderate or severe adverse events, infusion reactions and delayed reactions.

Table 4.   Adverse events in infliximab-controlled studies
ReferenceTotal patients with adverse events (%)Mild-Moderate adverse eventsSevere adverse eventsInfusion reactionsDelayed reactions
InfliximabPlaceboInfliximabPlaceboInfliximabPlaceboInfliximabPlaceboInfliximabPlacebo
  1. * Some patients may suffer from more than one adverse effect.

Järnerot et al.348*/24 (33)8*/21 (38)43112311
Probert et al.49No detailed data provided02No detailed data provided
Rutgeerts et al. (ACT 1)52–54217/243 (89)103/121 (85)4212345531271322
Rutgeerts et al. (ACT 2)54, 55195/241 (81)90/123 (73)2821372424281010
Sands et al.568*/8 (100)1/3 (33)70210000
 InfliximabSteroidsInfliximabSteroidsInfliximabSteroidsInfliximabSteroidsInfliximabSteroids
Armuzzi et al.20, 21No detailed data provided
Ochsenkuhn et al.47No detailed data provided18000000

Data synthesis

The main outcome considered in this review was percentage of response (defined by the authors of each study as partial or complete symptomatic response) and remission (defined by the authors as complete symptomatic response), both at the short-term (the first control performed in the study) and at the long-term (the last control performed in the study). The mean percentage of response/remission was calculated and expressed as weighted mean (and corresponding 95% confidence interval, 95% CI) to make due allowance for the number of patients included in each study.

For the meta-analysis, the homogeneity of effects throughout studies was appraised using a homogeneity test based on the chi-squared test. Because of the low power of this test, a minimum cut-off value of P =0.1 was established as a threshold of homogeneity, lower values indicating heterogeneity. In addition, the I2-statistic was calculated to assess the impact of heterogeneity on the results. This statistic describes the percentage of the variability in effect estimates that it is because of heterogeneity rather than sampling error (chance); a value >50% may be considered substantial heterogeneity.

Meta-analysis was performed combining the odds ratios (OR) of the individual studies in a global OR, using a fixed effect model (Peto method) when heterogeneity was not demonstrated, or a random effect model (DerSimonian and Laird) when results were heterogeneous. Significance and 95% CI were provided for the combined OR. In addition, ‘numbers needed to treat’ (NNT) with infliximab (an estimate of how many patients need to receive infliximab before one patient would experience response/remission, in comparison with placebo) were also provided. The OR for the occurrence of adverse events in the infliximab group compared with the control group was calculated. The ‘number needed to harm’ (NNH) was also derived from the tolerance data. All calculations were performed with the freeware program Review Manager 4.2.8., developed by the Cochrane Collaboration.

Subanalysis/sensitivity analysis

In the systematic review, subanalyses of efficacy with infliximab therapy were planned a priori depending on the age of the patients (children or adults), the indication for infliximab use (corticorefractory or corticodependent UC), the prescribed dose of infliximab (5 or 10 mg/kg), and the readministration of infliximab infusions during follow-up (after remission has been achieved). In the meta-analysis, subanalyses of efficacy with infliximab therapy were planned depending on the control group (placebo or steroids), the dose of infliximab (5 or 10 mg/kg) and the quality of the studies (based on quality score proposed by Jadad, see appropriate section).

Results

Description of studies

With the predefined search strategy, 521 articles were retrieved in MEDLINE, 750 in EMBASE, and two in CINAHL. All but one of the exclusion reasons were obvious, as the excluded studies clearly did not fulfil inclusion criteria, e.g. they did not evaluate efficacy of infliximab for the treatment of UC. However, one study was excluded because response with infliximab, although evaluated for the treatment of UC, was not separately provided for Crohn's disease and for UC patients.18 Finally, 34 studies, including a total of 896 patients, evaluating infliximab therapy in UC patients were included in the systematic review (Table 1).19–59

Severity of UC was clearly addressed in 322 of the included patients: of those, 207 (64.2%; 95% CI 59–70%) had severe disease, 110 (34.2%, 95% CI 29–39%) had moderate disease, and only five patients (1.6%, 0.7–3.6%) had mild disease. Fifty-two per cent (95% CI 49–55%) of the 877 patients in whom steroid-resistance status was established, were classified as being refractory to steroids.

Information about concomitant immunosuppressive therapy before starting infliximab infusion was provided in 694 patients: 318 (46%; 95% CI 42–50%) were taking these drugs, azathioprine or 6-mercaptopurine in most of the cases, while treatment with calcineurinic agents (cyclosporine or tacrolimus) was only rarely reported. On the other hand, the prescription of immunosuppressive agents after starting infliximab therapy was reported in 45 of the 98 (46%; 95% CI 36–56%) patients where this information was available.

Twenty percent of the patients in whom the exact number of infliximab infusions was described, received a single infusion as the initial infliximab induction regimen, although reinfusions at the physician's discretion was permitted in most (65%; 95% CI 62–68%) of the studies, including 746 patients (85% of the cases).

Mean follow-up at the short term (the first control performed in the study) was 2.3 weeks (range 0.14–8 weeks), and at the long term (the last control performed in the study) was 8.9 months (range 1.5–25.6 months).

Efficacy of infliximab treatment in UC

Mean (weighted mean) short-term response (partial or complete) and remission (only complete response) with infliximab therapy was 68% (95% CI 65–71%) and 40% (95% CI 36–44%) respectively. Respective figures for mean long-term response and remission with infliximab therapy was 53% (95% CI 49–56%) and 39% (95% CI 35–42%).

Efficacy of infliximab vs. placebo in UC

Five randomized studies, all of them double-blinded, compared infliximab with placebo for the treatment of UC (Table 2).34, 49, 52–56

As shown in Figure 1, short-term response was 65% (95% CI 61–69%) with infliximab, and 33% (95% CI 27–38%) with placebo. The OR (fixed effect model) for this effect was 3.6 (95% CI 2.67–4.95; P < 0.001). There was no statistically significant heterogeneity (test for heterogeneity chi-squared, P = 0.3; I2-statistic = 18%). NNT with infliximab to achieve short-term response was 3 (95% CI 3–4%). Short-term remission was achieved in 33% (95% CI 29–37%) of the patients treated with infliximab, and in 10% (95% CI 6.4–14%) of those receiving placebo. The corresponding OR (random effect model) was 4.56 (95% CI 1.98–10.5; P < 0.001), results being statistically heterogeneous (test for heterogeneity chi-squared, P = 0.09; I2-test = 66%). NNT with infliximab to achieve short-term remission was 4 (95% CI 3–6%).

Figure 1.

 Controlled studies comparing efficacy of infliximab 5 or 10 mg/kg vs. placebo.

Long-term response was 53% (95% CI 49–58%) with infliximab, and 24% (95% CI 19–29%) with placebo (Figure 1). The OR (fixed effect model) for this effect was 3.4 (95% CI 2.52–4.59; P < 0.001). There was no statistically significant heterogeneity (test for heterogeneity chi-squared, P = 0.76; I2-statistic = 0%). NNT with infliximab to achieve long-term response was 3 (95% CI 3–4). Finally, long-term remission was achieved in 33% (95% CI 29–37%) of the patients treated with infliximab, and in 14% (95% CI 9–18%) of those receiving placebo. The corresponding OR (random effect model) was 2.72 (95% CI 1.92–3.83; P < 0.001), and there was no statistically significant heterogeneity (test for heterogeneity chi-squared, P = 0.59; I2-statistic = 0%). NNT with infliximab to achieve long-term remission was 5 (95% CI 4–7).

Efficacy of infliximab vs. steroids in UC

Two randomized studies, both open-label trials, compared infliximab and steroids for the treatment of UC (Table 3).20, 21, 47 The low number of studies (only two, one of them classified as a low-quality study), and the low number of patients included in each study (10 and six patients respectively), prevented us from combining the results in a formal meta-analysis.

Subanalysis of efficacy

Patient's age

Ninety-two children were included in this systematic review, and infliximab achieved short-term (mean 2.2 weeks) response and remission in 75% (95% CI 64–83%) and 63% (95% CI 47–76%) of the cases respectively. Long-term (mean 7.9 months) response was 43% (95% CI 33–55%). Long-term remission (57%) could be calculated in only one study, which included only 14 children. On the other hand, 804 adults were treated with infliximab, achieving short-term (mean 2.6 weeks) response and remission in 68% (95% CI 65–72%) and 39% (95% CI 35–43%) of the cases respectively. Respective figures for long-term (mean 10.2 months) response and remission in adults were 54% (95% CI 50–57%) and 38% (95% CI 34–42%) respectively.

Indication for infliximab use

Similar response rates were calculated in steroid-refractory (70%; 95% CI 66–75%) and in non-steroid refractory patients (68%; 95% CI 63–72%) in the short term. The long-term response in steroid refractory patients was 56% (95% CI 50–63%), while the corresponding figure of efficacy with infliximab in steroid-dependent cases was provided in only 25 patients (84%; 95% CI 65–94%).

Infliximab dose

Infliximab was prescribed at 5 mg/kg dose in all but in four studies (Table 1). The dose of 10 mg/kg was administered to 245 patients, achieving short-term response and remission in 65% (95% CI 59–71%) and 30% (95% CI 24–36%) of the patients respectively. Respective figures for long-term response and remission in those who received infliximab 10 mg/kg, calculated from the ACT 1 and ACT 2 studies (the only ones with this information available), were 52% (95% CI 46–58%) and 35% (95% CI 29–41%). When subanalysis were performed for the meta-analysis of infliximab vs. placebo depending on the drug dose, infliximab at 5 mg/kg was more effective than placebo when all endpoints were considered: short-term response (OR 3.64; 95% CI 2.59–5.11), short-term remission (OR 5.28; 95% CI 2.3–12.1), long-term response (OR 2.92; 95% CI 2.05–4.16), and long-term remission (OR 2.61; 95% CI 1.69–4.03) (Figure 2). Results were homogeneous for all comparisons excepting for short-term remission, where borderline heterogeneity was detected (P = 0.11; I2-statistic = 60%). Finally, the infliximab dose of 10 mg/kg was also demonstrated to be more effective than placebo, with OR figures, respectively, for short-term response and remission of 3.61 (95% CI 2.54–5.15) and 3.59 (95% CI 2.52–5.1), and for long-term response and remission of 3.9 (95% CI 1.7–8.93) and 3.22 (95% CI 2.13–4.87) (Figure 3). Again, results were homogeneous for all comparisons excepting for short-term remission, where borderline heterogeneity was detected (P = 0.12; I2-statistic = 59%).

Figure 2.

 Controlled studies comparing efficacy of infliximab 5 mg/kg vs. placebo.

Figure 3.

 Controlled studies comparing efficacy of infliximab 10 mg/kg vs. placebo.

Infliximab reinfusions

In those studies where reinfusions at the physician's discretion was permitted, long-term response and remission to infliximab was 52% (95% CI 48–56%) and 37% (95% CI 34–41%) respectively. Nevertheless, these results should be interpreted with caution, as the statement that ‘reinfusions at the physician's discretion permitted’ does not necessarily indicate that all patients included in such a study received infliximab reinfusions. Only in a few studies infliximab reinfusions were not permitted, and therefore long-term response without reinfusions could be calculated in only 95 patients (56%; 95% CI 45–66%), and long-term remission in only 25 patients (68%; 48–83%).

Study quality

Our initial intention was to perform subanalysis depending on the quality of the study. However, regarding the meta-analysis of infliximab vs. placebo, all studies were of high quality, as all were randomized and double blind, with all studies but one having the highest quality according to the Jadad scale (Table 2). Therefore, the influence of this variable could not be adequately assessed.

Adverse events of infliximab vs. placebo in UC

Adverse events in infliximab-controlled studies are summarized in Table 4. Most of the studies reported only the number of adverse events; as a single patient may suffer more than one side-effect, in some cases it may not be possible to extract the exact number of patients with each adverse effect. Adverse effects were reported in 83% (95% CI 80–86%) of the infliximab-treated patients, and in 75% (95% CI 70–81%) in the placebo group. The OR (fixed effects model) for adverse events of infliximab vs. placebo was 1.52 (95% CI 1.03–2.24; P = 0.04; borderline heterogeneity: P = 0.12; I2-statistic = 48%) (Figure 4), and the NNH was 14 (95% CI 5–25).

Figure 4.

 Controlled studies comparing adverse effects of infliximab vs. placebo.

Discussion

Ulcerative colitis and Crohn's disease are distinct entities, but an imbalance between pro-inflammatory and anti-inflammatory cytokine production is suspected in both conditions.60 The cytokine profiles of UC and Crohn's disease are usually different.60 This is reflected by the different cytokines that the CD4+ T cells, T helper 1 (Th1), and T helper 2 (Th2) produce. In Crohn's disease, an over-expression of Th1-related pro-inflammatory cytokines, such as interferon gamma and TNF-α, has been classically considered.61 In contrast, UC has been traditionally associated with a dysregulated Th2 response and production of IL-4 and -5 resulting in the stimulation of humoral immunity.61 However, although the Th2-type of immune response is predominant in UC, it has been shown that TNF-α may also play a role in its pathogenesis. In this respect, patients with UC, on severe flare-up, show pathological and clinical aspects similar to those of Crohn's disease.62

The first line of evidence for the potential efficacy of anti-TNF-α therapy in the treatment of UC is the descriptive characterization of high levels of TNF-α production on either the local or systemic level.63 This includes the demonstration of an increased production of this cytokine in colonic mucosa, stools, rectal dialysate or plasma from patients with active UC; furthermore, the increased concentration correlates with disease severity, both clinically and endoscopically.64–70 The second line of evidence comes from animal data,63 as beneficial anti-TNF-α treatment effect has also been demonstrated in the cotton-top tamarin, an animal model of idiopathic UC.71

The first clinical trial evaluating the efficacy of anti-TNF-α treatment in patients with UC was performed in 1997 and used CDP 571, a chimeric monoclonal antibody directed against TNF-α.72 Thereafter, several studies have evaluated infliximab therapy in UC patients and have been included in the present systematic review (Table 1).19–59 Mean short-term response with infliximab therapy was 68%, thus suggesting that this drug is relatively effective for the treatment of UC, although the initial success achieved with this drug may diminish with time. Nevertheless, the efficacy of infliximab in UC is not as pronounced if one considers clinical ‘remission’– and not only ‘response’– as a reasonable endpoint for treatment,73 as only about 40% of the infliximab-treated patients will achieve this endpoint.

However, most of the aforementioned studies were uncontrolled. Therefore, controlled studies are necessary to definitively confirm these results. The first randomized double-blind, placebo-controlled trial using infliximab in severe steroid-refractory UC was conducted by Sands et al. in 2001.56 In this study, 11 patients with UC having active disease and receiving at least 5 days of intravenous steroids were randomized to receive a single intravenous infusion of infliximab or placebo. Four of eight patients (50%) who received infliximab were considered treatment successes at 2 weeks, compared with none of three patients who received placebo. The second randomized, double-blind, placebo-controlled trial was performed by Probert et al.,49 also in steroid-resistant UC. Infliximab infusions of 5 mg/kg were given at weeks 0 and 2. After 2 weeks, there was no statistically significant difference between the infliximab and placebo groups in the proportion of patients with response (13% vs. 5%). After 6 weeks, remission rates were 39% vs. 30%. The authors concluded that these data do not support the use of infliximab in the management of moderately active steroid-resistant UC, although the number of patients was quite low and therefore this study was underpowered to demonstrate statistically significant differences between the placebo and the infliximab groups.

Traditionally, acute attacks of UC have been treated with corticosteroids intravenously. However, despite this intensive treatment, severe attacks have a high colectomy rate.4, 74 In this context, Jarnerot et al.34 conducted the third randomized double-blind placebo-controlled trial of infliximab in 45 severe to moderately severe UC patients not responding to conventional treatment with steroids. The primary endpoint was colectomy or death 3 months after randomization. Seven patients in the infliximab group and 14 in the placebo group had a colectomy (a statistically significant difference) within 3 months after randomization.

Finally, and most recently, a multicentre, international, randomized, placebo-controlled trial, named ACT 1, was designed to evaluate the safety and efficacy of infliximab for UC.52–54 Patients (364) with active UC despite the use of corticosteroids/azathioprine/6-mercaptopurine, with endoscopic evidence of moderate or severe UC and a total Mayo score of 6–12, were randomized to receive placebo, infliximab 5 mg/kg or 10 mg/kg at weeks 0, 2 and 6 weeks and thereafter every 8 weeks through week 46. Significantly higher proportions of patients receiving infliximab were in clinical response vs. placebo-treated patients at week 8 (65% vs. 37%) and at week 54 (51% vs. 20%). From a more exigent point of view, 35% of infliximab-treated patients were in clinical remission (and not only ‘response’) vs. 16% of placebo-treated patients at week 8, and these differences in remission rates persisted at week 54 (35% vs. 16%, placebo. In a similarly designed study, the ACT 2 trial, 364 patients with UC, refractory to at least one standard therapy including 5-aminosalycilates (this was the main difference with respect to ACT 1 trial), corticosteroids or immunosuppressants, were also randomized to receive infliximab 5, 10 mg/kg, or placebo at weeks 0, 2, 6, 14 and 22.54, 55 Clinical response with infliximab vs. placebo was 57% vs. 30% at week 8, and 54% vs. 26% at week 30. Remission rates were also higher in the infliximab group, both at week 8 (35% vs. 5.7%) and at week 30 (35% vs. 11%).

In summary, five randomized, double-blinded and high-quality studies, have compared infliximab with placebo for the treatment of UC (Table 2).34, 49, 52–56 As shown in Figure 1, our meta-analysis showed an advantage of infliximab over placebo in all endpoints considered (short- and long-term response and remission), with ORs ranging from 2.7 to 4.6, and NNTs ranging from 3 to 5 only. NNTs are specific to a particular length of follow-up as they are based on the number of people who will benefit within a certain period of time who otherwise would not benefit. Systematic reviews tend to combine trials of varying follow-up periods, which could make an NNT difficult to interpret, and therefore NNTs should be taken with caution when the follow-up periods are different. Nevertheless, almost all the patients included in our meta-analysis came from the ACT trials and, therefore, when NNTs were calculated separately from only ACT 1 and ACT 2 studies, results were similar (NNT from 4 to 5).

In addition, two randomized studies, both open-label trials, have compared infliximab and steroids for the treatment of UC. Armuzzi et al.20, 21 concluded that infliximab was as effective as steroids in the management of moderate-to-severe steroid-dependent UC. Ochsenkuhn et al.47 evaluated whether infliximab could achieve remission in patients with acute severe UC who were not yet steroid-refractory, and their results suggested that infliximab could be used in this setting as a substitute therapy in patients for whom steroids are contraindicated.47 In summary, two randomized studies have compared infliximab and steroids for the treatment of UC (Table 3).20, 21, 47 However, the low number of studies (only two, one of them a low-quality study), and the low number of patients included in each study, prevented us from combining the results in a formal meta-analysis.

Several variables may influence the efficacy of infliximab therapy in UC. With respect to patient's age, one study has suggested that paediatric patients may respond better than adults.75 However, only a few authors have included in the same study both children and adults. One of these studies found that the factors influencing on infliximab response were similar in both high and low age groups.41 Others, including only adults, showed that age presents no influence on infliximab response.42 Our systematic review showed some differences in response to infliximab depending on the age. However, these differences were sometimes contradictory (e.g. higher short-term results in children, but better long-term results in adults). Furthermore, these results should be interpreted with caution, as other relevant variables (with potential influence on infliximab efficacy, such as follow-up, indication for infliximab use, severity of the disease, concomitant immunosuppresive therapy or infliximab dose) were not equally distributed in children and adults.

The indication for infliximab use (that is, steroid-dependent vs. steroid-refractory UC) may also influence the efficacy results. Some studies have achieved higher response rates in steroid-refractory than in steroid-dependent cases,41, 51 suggesting that patients with more acute disease secreting increased levels of TNF-α may be more responsive to therapy.76 In this respect, in the ACT 1 trial, clinical response at week 8 was 73% in corticosteroid-refractory subjects, while it was somewhat lower, 63%, in non-corticosteroid-refractory cases.52–54 However, in the ACT 2 trial, these differences were not confirmed.54, 55 Furthermore, other authors have even reported opposite results, with higher response rates in steroid-dependent than in steroid-refractory cases.27, 29, 42, 59, 75 In summary, subanalysis from our review allowed to calculate similar response rates in steroid-refractory and in non-steroid-refractory patients, suggesting that, probably, infliximab may be used for both indications.

The currently recommended dose of infliximab for treatment of patients with Crohn's disease is 5 mg/kg,8–10 and, accordingly, infliximab has been prescribed at 5 mg/kg dose in all but in four UC studies (Table 1). However, it has been suggested that, in Crohn's disease, 10 mg/kg every 8 weeks may be superior to 5 mg/kg every 8 weeks for maintenance therapy.77 In the ACT 1 trial,52–54 clinical response and remission was similar with infliximab 5 and 10 mg/kg both at the short-term and at the long-term follow-up. In the ACT 2 trial,54, 55 clinical response and remission rates with both infliximab dose schemes were similar at 8 weeks. However, at the end of the study (30 weeks), a slight advantage of 10 mg/kg over 5 mg/kg was observed, both in the clinical response rate (60% vs. 47%) and in the remission rate (36% vs. 26%). When we performed subanalysis for the meta-analysis of infliximab vs. placebo depending on the drug dose, both infliximab 5 mg/kg (Figure 2) and 10 mg/kg (Figure 3) were more effective than placebo, with roughly similar ORs. Therefore, it may be suggested that the preferred initial dose of infliximab in patients with UC is 5 mg/kg.54

In patients with Crohn's disease, a significantly better initial response and a more prolonged response has been reported, in some studies, in those taking concurrent immunosuppressive medications.78 In this same way, some authors have reported a better response to infliximab or a lower rate of relapses in patients with UC and concomitant use of immunosuppressive agents such as azathioprine or 6-mercaptopurine.29, 32 However, others have shown that these drugs do not affect the short- or the long-term outcome in UC, as was the case in the ACT 1 and ACT 2 trials.52–55 Therefore, the role of anti-metabolites in combination with infliximab for the treatment of UC will need to be clarified.

Cyclosporine constitutes an efficacious alternative in the treatment of severe, steroid-refractory, attack of UC. A recent systematic review concluded that intravenous cyclosporine achieved remission in 91% and 71% of patients with a severe attack of UC, in controlled and uncontrolled studies.7 In other reviews, a 70–80% initial response of UC patients to cyclosporine was calculated.6, 79, 80 Thus, short-term response rates with infliximab in corticosteroid-refractory UC patients seem to be inferior to those reported with intravenous cyclosporine, although the two drugs have not been compared directly. However, the failure to attain long-term response with infliximab seems to be similar to the pattern observed with cyclosporine, as the likelihood of avoiding colectomy over 2–3 years is only about 50% or less in UC patients previously treated with this last drug.6, 79, 80

As no study up to now has compared infliximab with cyclosporine for corticosteroid-resistant UC, the question of which drug should be used first remains unanswered. Therefore, other factors will have to be taken into consideration, most importantly the side-effect profile. Cyclosporine has well-recognized important side-effects such as hypertension, nephrotoxicity, seizures and infections, with several deaths reported.5–7 Infliximab may also be associated with the development of adverse effects.81 In the present meta-analysis, adverse effects were reported in 83% of the infliximab-treated patients, and in 75% of those receiving placebo. It is remarkable the high incidence of adverse effects related to infliximab, although most of them were mild (Table 4). The incidence of severe adverse events was rarely reported and very similar in the infliximab and placebo groups. It is also noteworthy the high and unexpected rate of adverse effects with placebo, which underlines the fact that randomized, placebo-controlled trials are the most objective means of evaluating not only drug efficacy, but also safety.82 Nevertheless, the NNH (for adverse events of infliximab vs. placebo) was 14, a relatively high figure, which favourably compares with the aforementioned NNT of only 4 approximately. In summary, although the short-term response rates with infliximab in corticosteroid-refractory UC patients seem to be somewhat inferior to those reported with cyclosporine, the better safety profile and the simplicity of infliximab (cyclosporine requires initial continuous parenteral administration and close monitoring to maintain therapeutic blood levels) suggest that infliximab may be an alternative therapy in patients who are being considered for cyclosporine therapy or colectomy. In addition, infliximab can be considered in patients with acute steroid-refractory disease who are reluctant to undergo colectomy and in whom cyclosporine is contraindicated.73 Nevertheless, more studies are necessary to evaluate the cost–benefit ratio of infliximab use, especially when compared with cyclosporine, in corticosteroid-refractory UC patients.

A limitation of the present systematic review is that it is based, in a great part, on many small studies with relevant methodological differences and, consequently, sometimes contradictory results. The reasons for the discrepancies among studies are not clear, but comparison among published series is difficult, as the protocols markedly differ. One explanation for the conflicting results obtained with infliximab in different studies in UC might be heterogeneity of patients treated (see Table 1). For example, some studies include paediatric patients, while others treat only adults. The spectrum of severity of UC also varies among studies, with some protocols including only severe cases, and others including mild or moderate patients.83 Furthermore, inclusion criteria for some studies are UC patients refractory to steroids or immunosuppresants, while in others the refractoriness to 5-aminosalycilates (as in the ACT 2 trial) was sufficient to be included. In addition, the proportion of steroid-dependent and steroid-refractory cases also differs among studies. The previous use and the concurrent prescription of azathioprine or 6-mercaptopurine may further influence the response to infliximab in some studies. Although the dose of infliximab has been 5 mg/kg in most studies, some protocols have prescribed 10 mg/kg or even higher doses. Furthermore, the number of infusions and the maintenance regimens (e.g. infusions every 8 weeks) are not homogeneously prescribed. In fact, in most open-label trials, infliximab was given at the discretion of the gastroenterologist with no uniform schedule of infusions. Differences in the definition of clinical response or remission may also explain, at least partly, some of the differences among studies. In this respect, the considered outcome measures, defined by the authors of each study, varied among the different protocols (Table 5). Nevertheless, in all of the controlled studies, well-defined activity indexes were employed to evaluate the severity of the disease. All of them are capable of properly identifying patients in clinical remission, and therefore we believe that the results obtained using these different scores may be comparable. Finally, publication bias may partly explain reports of relatively favourable trends associated with infliximab treatment in UC as negative results in one or a few patients are unlikely to be published. Nevertheless, it may be concluded, mainly based on the results of our meta-analysis, that infliximab is more effective than placebo – for the treatment of moderate-to-severe UC. However, data on the long-term efficacy of infliximab are still lacking, underlying the fact that many unanswered questions still remain which can only be answered by further investigation.

Table 5.   Outcome measures used in infliximab-controlled studies
ReferencePrimary outcomeEndoscopic evaluation
Järnerot et al.34Colectomy
Probert et al.49Ulcerative colitis score systemBaron score
Rutgeerts et al. (ACT 1)52–54Mayo ScoreMayo subscore for endoscopy
Rutgeerts et al. (ACT 2)54, 55Mayo ScoreMayo subscore for endoscopy
Sands et al.56Modified Truelove–Witts indexBlackstone classification
Armuzzi et al.20, 21Disease activity indexNo
Ochsenkuhn et al.47Modified Truelove–Witts indexNot specified

Acknowledgement

Supported in part by Grants from the Instituto de Salud Carlos III (C03/02 and PI050109). This study was not funded by any Pharmaceutical Company. The authors state they have no conflict of interest.

Ancillary