Methodological assessment was done using the Evidence-Based Gastroenterology Steering Group recommendations. ¹⁰ A Jaded score of each trial was also calculated. If a significant difference in methodological quality among studies was observed, a sensitivity analysis would be undertaken by excluding those trials with less quality. If relevant data was missing in original manuscripts, authors would be contacted.

Meta-analysis was performed using REVMAN software (Review Manager Version 5.2. Copenhagen: The Nordic Cochrane Collaboration, 2012). Heterogeneity among studies was evaluated by means of chi square and I2 tests. A random-effect model was used to give a more conservative estimate of the effect of individual therapies, allowing for any heterogeneity among studies. Outcome measures were described as relative risk (RR) of failure to achieve remission and RR of relapse of disease activity in CD patients as well as in UC patients. Also, 95% confidence intervals were calculated. Funnel plots were designed to evaluate possible publication bias. Numbers necessary to treat (NNT) were calculated.

Two trials evaluating Vedolizumab efficacy on 555 patients ^{16, 17} and one phase II trials assessing the efficacy of Etrolizumab18 were included. The trial published by Feagan et al in 2013 included a cohort of patients enrolled in an open-label group, which was not considered for analysis. No trials on Natalizumab were found on this particular subject. No significant heterogeneity among trials was found. Results are shown in Figure 4. Vedolizumab showed a significantly lower RR of failure to induce remission versus placebo [RR 0.85 (0.77-0.94)], with a NNT of 8. Both trials assessed mucosal healing as an endpoint, though they used different scores (modified Baron Score and Mayo Score). When assessing Etrolizumab, the study by Vermeire et al showed that compared to placebo, Etrolizumab was significantly more effective in inducing clinical remission at week 10; however, more evidence is needed before drawing a valid conclusion since only a phase I and a phase II trial using Etrolizumab were published so far. Pooled results of Vedolizumab and Etrolizumab, as shown in Figure 4, still showed a significant efficacy of anti-integrin antibodies to induce remission. Once again, Vedolizumab showed a significantly lower RR of failure to induce mucosal healing versus placebo [RR 0.84 (0.74-0.94)]. Results are shown in Figure 5.

Figure 4. Figure 4. Forest plot of randomized controlled trials of anti-integrin antibodies versus placebo in inducing remission in active UC.

Figure 5. Figure 5. Forest plot of randomized controlled trials of anti-integrin antibodies versus placebo in inducing mucosal healing in active UC.

Tables 5 and 6 show adverse events incidence in patients treated with Natalizumab and Vedolizumab respectively versus placebo. Overall, no significant differences were found in terms of serious adverse events or serious infections when comparing Natalizumab versus placebo. Progressive Multifocal Leukoencephalopathy (PML) was not reported in any of the included studies. On the other hand, Vedolizumab showed an increased risk in serious adverse events and serious infections.

Table 5. Table 5. Adverse events with Natalizumab vs. placebo in inducing remission in active CD.

Table 6. Table 6. Adverse events with Vedolizumab vs. placebo in inducing remission in active CD and UC.

As specified by the results of our meta-analysis, there is a class-effect of anti-integrin antibodies for remission induction in patients with CD and UC. There is less evidence supporting their role for remission maintenance. According to our knowledge, this systematic review puts on perspective the utility of this kind of drugs suggesting that in the future new medicines that interfere leukocyte adhesion may represent a valid therapeutic alternative.

There is growing evidence regarding anti-TNF antibodies efficacy for induction and maintenance of clinical remission in CD and UC. ²¹ However, a significant proportion of patients may not have a clinical response or lose response over time. ²² Thus, there is a need for new therapeutic strategies in this area.

Integrin a4 inhibition has a proven effect on inflammatory response, not only on intestinal but also on extra-intestinal inflammation. ⁸ Natalizumab inhibits a4ß7 and a1b1 integrins and is effective for CD treatment, as shown in previous meta-analysis by Ford et al. As a result of a1b1 inhibition, it has a significant effect on leukocyte adhesion in the central nervous system. In fact, it has been used as a therapeutic option for central nervous system autoimmune conditions, such as multiple sclerosis. ²⁴

Nevertheless, during the last few years, increasing reports on the development of PML caused by activation of JC virus in patients treated with Natalizumab have been published. ²⁵ As a consequence, its use has been restricted, resulting in the lack of further published experiences with this drug and the need for selective inhibition of a4b7 integrin. ²⁶ While Ertrolizumab ²⁷ and Vedolizumab have been developed, evidence of efficacy is strongest for Vedolizumab.

Vedolizumab is an IgG1 human antibody directed against a4b7 integrin that is effective for CD and UC in multicenter trials. ^{17, 19, 20} According to our meta-analysis, Vedolizumab seems to have a better performance on UC than CD. More evidence is still required to determine the real magnitude of these differences. What is more, additional evidence is still required to evaluate the efficacy of this kind of drugs for maintenance of remission. It is worth mentioning that only one trial evaluated mucosal healing as an outcome measure. ¹⁷ As a consequence, more evidence is still necessary on the efficacy of these drugs to induce and maintain mucosal healing. According to the trial published by Sands et al, Vedolizumab efficacy was greater when clinical outcomes were assessed at 10 weeks, after finishing induction therapy. ²⁰ This is an important point to be considered when conducting future clinical trials with anti-integrin antibodies, since remission should then be monitored a few weeks after induction therapy completion.

A rather promising therapeutic agent was also introduced in this systematic review: Etrolizumab, a humanized monoclonal antibody that selectively binds the b7 subunit of the heterodimeric integrins a4b7 and aEb7. There is very little evidence on its clinical efficacy, so caution must be taken when analyzing the results of the phase II clinical trial included in this systematic review.

Included trials with Vedolizumab have not shown any reported cases of PML. However, it is noteworthy that, unlike Natalizumab, Vedolizumab patients had a significant higher risk of serious adverse events and serious infections. This aspect highlights the need for further evidence assessing this aspect.

Finally, little evidence is shown on the efficacy of these drugs in patients who have already experienced TNF antagonist failure. Although anti-integrin antibodies would seem like a valid option in this clinical scenario, more evidence is still needed from prospective clinical trials. ²⁸

The main strength of this meta-analysis is that, it evaluates the class-effect of anti-leukocyte adhesion antibodies, instead of the individual effect of a single drug. This is relevant because it enforces the potential utility of future anti-leukocyte adhesion antibodies. Despite the rigorous search strategy, funnel plot asymmetry suggests the presence of publication bias that could be a significant limitation of this meta-analysis. We think, however, that this asymmetry may be due to the relatively scarce number of published trials. It is important to highlight that the results were expressed as a relative risk of failure to induce and/or maintain remission due to previously published meta-analyses such as the one published by Ford et al that use the same methodology. ²³

In conclusion, anti-integrin antibodies have shown a beneficial class effect for induction of clinical remission in patients with CD and UC. There is a need for more evidence on their efficacy for maintaining remission. Their role on IBD treatment still needs to be determined.

Authors’ contributions. JL performed the bibliographic search, undertook statistical analysis, contributed to the writing of the draft. AR performed the bibliographic search, contributed to the writing of the draft. IZ performed the bibliographic search, contributed to the writing of the draft, and reviewed the final draft.

Conflicts of interest. Astrid Rausch is consultant physician for Takeda Pharmaceuticals. Ignacio Zubiaurre and Astrid Rausch are speakers for Abbvie Pharmaceutical Company. Ignacio Zubiaurre is consultant physician for Ferring Pharmaceuticals. Juan Lasa declares no conflict of interests.