Medical Management Following Surgical Therapy in Inflammatory Bowel Disease: Evidence from Cochrane Reviews

INTRODUCTION

Although medical therapy for inflammatory bowel disease (IBD) has greatly improved over the past 2 decades, surgery continues to be an integral part of the management of ulcerative colitis (UC) and Crohn’s disease (CD). This article reviews data from Cochrane reviews relevant to the medical management of IBD in surgical patients including treatment and prevention of pouchitis in UC patients with an ileal pouch-anal anastomosis, the risk of postoperative infectious complications from medical therapies, infertility after IBD surgery, and prevention of recurrence after bowel resection for CD. A summary of the overall certainty of evidence was assessed using Grading of Recommendations Assessment (GRADE; Table 1), and a summary of the results of these reviews is provided in Table 2.

TREATMENT AND PREVENTION OF POUCHITIS AFTER ILEAL POUCH-ANAL ANASTOMOSIS FOR ULCERATIVE COLITIS

Pouchitis is an chronic inflammatory disease that may occur in the ileal pouch after proctocolectomy in patients undergoing ileal pouch-anal anastomosis (IPAA).¹ The risk of pouchitis is substantially higher in patients undergoing IPAA for UC compared with patients undergoing this procedure for familial adenomatous polyposis. Most patients present with a combination of symptoms including increased stool frequency, abdominal cramping, tenesmus, fecal urgency, and incontinence. A clinical diagnosis should be confirmed by endoscopy and biopsy of the pouch. Patients with pouchitis can be classified according to disease activity and symptom duration.^{2, 3} Disease activity can be classified as remission (ie, no active pouchitis), mild to moderately active disease (ie, increased stool frequency, urgency, infrequent incontinence), or severely active disease (ie, hospitalization for dehydration, frequent incontinence). The Pouchitis Disease Activity Index (PDAI) is a 19-point composite index of pouchitis activity based upon clinical symptoms, endoscopy, and histology findings.⁴ Active pouchitis is defined as a PDAI ≥7, with remission defined as a PDAI <7. Clinical response to treatment is based on a reduction in the PDAI score ≥3 from baseline. It should be noted that the PDAI was developed empirically and has not been rigorously validated. Symptom duration can be classified as acute (ie, ≤4 weeks duration) or chronic (ie, >4 weeks duration).

A Cochrane review conducted by Nguyen et al included 15 randomized controlled trials (RCTs; N = 547) that evaluated the treatment of active pouchitis or the prevention of pouchitis after IPAA for UC.⁵ Several different interventions were assessed including antibiotics, probiotics, budesonide, adalimumab, bismuth carbomer foam enema, and allopurinol.

Efficacy of Interventions for Acute Pouchitis

Four RCTs evaluated the efficacy of medical interventions for treatment of acute pouchitis. The overall body of evidence in support of any intervention over another treatment or placebo was deemed to be of very low quality. Based on one small RCT of 16 patients, ciprofloxacin (1000 mg/d) may be more effective than metronidazole (20 mg/kg/d) for induction of remission at 2 weeks (100% [7 of 7] vs 33% [3 of 9]; risk ratio [RR], 2.68; 95% confidence interval [CI], 1.13–6.35).⁶ Due to a high risk of bias and a very small number of events generated by this small clinical trial, the certainty of evidence was rated as very low.

A second double-blind, double-dummy RCT that studied 26 patients with acute pouchitis compared the relative efficacy of oral metronidazole (0.5 g twice daily) to budesonide enemas (2 mg/100 mL daily) for induction of remission.⁷ No statistically significant differences in clinical remission or clinical response rates were found between metronidazole and budesonide. At 6 weeks, 43% (6 of 14) of patients in the metronidazole group achieved clinical remission, in comparison with 50% (6 of 12) of budesonide patients (RR, 0.86; 95% CI, 0.37–1.96). Clinical response was achieved by 50% (7 of 14) of metronidazole patients, in comparison with 58% (7 of 12) of budesonide patients (RR, 0.86; 95% CI, 0.42–1.74). The GRADE analysis determined the certainty of evidence supporting these outcomes was very low because of an unclear risk of bias and the small number of patients evaluated.

Another small placebo-controlled study that evaluated rifaximin (400 mg) induction therapy showed a statistically nonsignificant difference in remission rates (25% [2 of 8] vs 0% [0 of 10]; RR, 6.11; 95% CI, 0.33–111.71).⁸ Finally, the probiotic Lactobacillus GG (0.5–10) × 10¹⁰ colony-forming units/capsule twice daily was compared with placebo in a small RCT for induction of remission with equivocal results (10% [1 of 10] vs 0% [0 of 10]; RR, 3.00; 95% CI, 0.14–65.90).⁹ The certainty of evidence for both the rifaximin and Lactobacillus GG outcomes was rated as very low due to unclear risk of bias and very limited data.

Efficacy of Interventions for Chronic Pouchitis

Five RCTs were identified that evaluated the efficacy of interventions for maintaining remission in patients with chronic pouchitis. Based upon 2 RCTs, VSL#3 formulation (6 g/day) may be superior to placebo for maintaining remission at 9 to 12 months (85% [34 of 40] vs 3% [1 of 36]; RR, 20.24; 95% CI, 4.28–95.81).^{10, 11} Despite the very strong summary efficacy estimate generated using the aggregate data from these 2 trials, the certainty of evidence supporting this finding was low due to a limited amount of data.

Results from a small trial suggested that glutamine suppositories (1 g twice daily) may be more effective than butyrate suppositories (40 mmol twice daily) for induction of remission at 3 weeks (60% [6 of 10] vs 33% [3 of 9]; RR, 1.80; 95% CI, 0.63–5.16).¹² The certainty of evidence was determined to be very low because of unclear risk of bias and very limited data. Another RCT that evaluated topical maintenance therapy found no benefit of bismuth carbomer foam enema (metallic bismuth 270 mg) over placebo for clinical improvement at 3 weeks (45% [9 of 20] vs 45% [9 of 20]; RR, 1.00; 95% CI, 0.50–1.98).¹³ However, the certainty of evidence supporting this outcome was very low due to unclear risk of bias and very limited data.

Finally, in a small RCT that was terminated prematurely, the benefit of adalimumab (standard induction and maintenance dose of 160 mg, followed by 80 mg, followed by 40 mg every fortnightly) over placebo for clinical improvement at 4 weeks in patients with chronic, antibiotic-refractory pouchitis was equivocal (50% [3 of 6] vs 43% [3 of 7]; RR, 1.17; 95% CI, 0.36–3.76).¹⁴ The certainty of evidence supporting this outcome was rated as low due to very limited data.

Interventions for Prevention of Pouchitis

Six RCTs evaluated the efficacy of interventions for prevention of pouchitis after IPAA. The efficacy of VSL#3 in comparison with placebo or a “no treatment” control group was assessed in 2 RCTs, evaluating the proportion of patients with no episodes of pouchitis 12 months after IPAA. In the first trial, 90% (18 of 20) of patients receiving VSL#3 (1 packet per day) did not experience any episodes of pouchitis at 12 months compared with 60% (12 of 20) of placebo patients (RR, 1.50; 95% CI, 1.02–2.21).¹⁵ The results of the second trial were also favorable for VSL#3 (2 packets once per day): 100% (16 of 16) of VSL#3 patients did not experience any episodes of pouchitis at 12 months in comparison with 92% (11 of 12) of the no treatment control group (RR, 1.10; 95% 0.89–1.36).¹⁶ A GRADE analysis indicated that the certainty of evidence was low due to very limited data for the placebo-controlled study and very low due to high risk of bias and very limited data for the study with a no treatment comparison group.

The efficacy of another probiotic formulation, Bifidobacterium longum, over placebo for preventing pouchitis was evaluated in a single study that yielded uncertain results. At 6 months, 86% (6 of 7) of patients in the probiotics group did not experience any episodes of pouchitis in comparison with 60% (3 of 5) of the placebo group (RR, 1.43; 95% CI, 0.66–3.11; very low certainty evidence).¹⁷ Likewise, a second trial that evaluated a Clostridium butyricum MIYAIRI formulation (20 mg, 3 tabs daily) also generated inconclusive results. At 24 months, 11% (1 of 9) of the probiotics group did not experience any episodes of pouchitis in comparison with 50% (4 of 8) of the placebo group (RR, 0.22; 95% CI, 0.03–1.60; very low certainty evidence).¹⁸

The efficacy of tinidazole (500 mg daily) was compared with placebo in a single study. At 12 months, 81% (21 of 26) of patients in the tinidazole group did not experience any episodes of pouchitis compared with 58% (7 of 12) of placebo patients (RR, 1.38; 95% CI, 0.83–2.31; very low certainty evidence).¹⁹ Finally, allopurinol (100 mg twice daily) was compared with placebo in a single study. At 24 months, 46% (43 of 94) of allopurinol patients did not develop pouchitis compared with 43% (39 of 90) of the placebo group (RR, 1.06; 95% CI, 0.76–1.46; low certainty evidence).²⁰ The certainty of evidence for tinidazole and allopurinol were both rated as very low due to unclear risk of bias and very limited data.

In summary, the currently available data show that the efficacy of antibiotics, probiotics, and other pouchitis therapies are largely uncertain for both induction and maintenance therapy. Although clinical experience indicates that broad spectrum antibiotics are an appropriate first-line treatment strategy for pouchitis, multiple unmet needs exist. Specifically, adequately powered and well-designed trials are required to establish the best therapy for both induction of remission and prevention of pouchitis.

RISK OF POSTOPERATIVE INFECTIOUS COMPLICATIONS FROM MEDICAL THERAPIES IN INFLAMMATORY BOWEL DISEASE

Many medications used to treat IBD may be associated with an increased risk of infection due to their immunosuppressive effects.²¹ Furthermore, patients with IBD commonly undergo surgical procedures, both for management of their disease or other conditions. The immunosuppressive and anti-inflammatory effects of IBD therapy raise concerns regarding whether use of these agents in the perioperative period convey an increased risk of postoperative complications, including infections. As a result of these fears, some surgeons recommend patients discontinue IBD medications for several weeks before undergoing elective surgery.²² However, this strategy may be associated with other risks, including the reactivation of IBD. For patients using biologics, a drug holiday could also lead to immunogenicity and loss of response to the biologic after resumption of treatment postoperatively. Studies evaluating the association between perioperative medications and postoperative infections have yielded mixed results, particularly the studies examining risk of postoperative infections from perioperative tumor necrosis factor-alpha (TNF-α) antagonists.^23–27

A systematic review conducted by Law et al evaluated the risk of postoperative infectious complications in patients using common conventional and biologic therapies for IBD.²⁸ Given that no appropriate RCTs have been conducted in this field, only observational studies were included in the review. A total of 63 studies satisfied the inclusion criteria for meta-analysis. The primary outcome was postoperative infectious complications within 30 days of surgery. The secondary outcome was intra-abdominal infectious complications. Prespecified subgroup analyses reported on whether type of IBD or year of publication (before 1998 vs after 1998—the year infliximab was approved by the USFDA) influenced the estimates assessed.

There were 35 eligible studies that examined the risk of infection with preoperative corticosteroid use. Overall, patients exposed to preoperative corticosteroids had an increased risk of developing a postoperative infectious complication compared with those not exposed to corticosteroids (odds ratio [OR], 1.34; 95% CI, 1.25–1.44). Subgroup analysis by type of IBD showed a similar risk of postoperative infectious complications in patients with CD (OR, 1.27; 95% CI, 1.14–1.40) and UC (OR, 1.37; 95% CI, 1.22–1.53). Similarly, a significantly increased risk of postoperative infectious complications was seen in studies performed before 1998 (OR, 1.74; 95% CI, 1.26–2.41) and after 1998 (OR, 1.32; 95% CI, 1.23–1.42). A secondary analysis that specifically evaluated intra-abdominal infection also showed an increased risk with corticosteroid use (OR, 1.63; 95% CI, 1.33–2.00).

There were 5 eligible studies that examined the risk of infection with preoperative 5‐aminosalicylates (5-ASA) use. Patients exposed to 5-ASA formulations preoperatively had a significantly lower risk of infectious complications than unexposed patients (OR, 0.63; 95% CI, 0.46–0.87). There was no statistically significant difference in the risk of postoperative infectious complications for studies conducted before 1998 (OR, 1.08; 95% CI, 0.47–2.51). However for studies conducted after 1998, patients exposed to 5-ASA had a significantly decreased risk of postoperative infectious complications compared with unexposed patients (OR, 0.57; 95% CI, 0.40–0.81).

A total of 26 eligible studies evaluated the risk of postoperative infectious complications in patients receiving immunosuppressives including thiopurines, methotrexate, cyclosporine, and tacrolimus. There was no statistically significant difference in postoperative infectious complications between those exposed to preoperative immunosuppressives compared with those who did not receive immunosuppressives (OR, 1.08; 95% CI, 0.94–1.25). No difference in postoperative infection risk was seen in patients with CD (OR, 1.06; 95% CI, 0.83–1.36) or those with UC (OR, 1.07; 95% CI, 0.83–1.39). In contrast, studies published before 1998 demonstrated an increased risk of postoperative complications in patients treated with these agents (OR, 1.85; 95% CI, 1.14–3.01); however, this observation was not seen in studies published after 1998 (OR, 1.03; 95% CI, 0.88–1.20). Before the introduction of biologics, immunosuppressives were used more often for those with severe disease but are now mainly reserved for patients with mild to moderate disease. Hence, the increased risk of postoperative infection in studies performed before 1998 may reflect confounding by disease severity rather than a true biological effect.

A total of 49 eligible studies examined postoperative infection risk in patients exposed to TNF-α antagonists. Overall, an increased risk of postoperative infectious complications was observed in patients exposed to preoperative anti-TNF-α antagonists (OR, 1.26; 95% CI, 1.07–1.50). On subgroup analysis, the excess risk was restricted to patients with CD (OR, 1.48; 95% CI, 1.11–1.97) and was not observed in patients with UC (OR, 1.05; 95% CI, 0.79–1.41). In a post hoc subgroup analysis that examined patients who received TNF-α antagonist therapy within 8 weeks of surgery, a significantly elevated risk of postoperative infection was observed (OR, 1.44; 95% CI, 1.08–1.93). For patients who received TNF-α antagonist therapy greater than 8 weeks before surgery, the risk of postoperative infection was no longer statistically significant (OR, 1.15; 95% CI, 0.93–1.43).

There were 8 eligible studies that assessed the risk of postoperative infectious complications from preoperative anti-integrin (eg, natalizumab or vedolizumab) use. In the overall analysis, no attributable risk of infection was identified in those using anti-integrin therapies compared with patients who were not exposed to anti-integrin therapies (OR, 1.06; 95% CI, 0.67–1.69). Only 1 study examined the risk of postoperative infection in patients treated with anti-interleukin therapies (eg, ustekinumab) and found no significant difference in postoperative infections in patients using these therapies compared with unexposed patients (OR, 0.98; 95% CI, 0.58–1.66).²⁹

The collective quality of evidence in these studies was very low, largely due to the observational nature of the studies included. There was a serious risk of bias detected, and high levels of imprecision in the results. Notably, the majority of these studies were unable to adequately adjust for known potential confounders, most critically disease activity and severity. The authors concluded that although the meta-analysis revealed an increase in the odds of postoperative infections in patients using preoperative corticosteroids and TNF-α antagonists, in addition to a decrease in postoperative infection risk in patients using preoperative 5-ASA therapies, it is difficult to draw firm conclusions from this study due to the presence of residual confounding. This issue likely explains the differences observed when studies conducted before 1998 were compared with those conducted after 1998. Before 1998 when no biologics were available, patients with more severe disease tended to use preoperative immunosuppressives. In studies conducted before 1998, a higher risk of postoperative infectious complications was found in patients using preoperative immunosuppressives. After 1998, patients with severe disease were more likely to be treated with biologics and accordingly were less likely to receive immunosuppressives such as azathioprine (AZA). Likewise, patients with mild disease activity and/or a good prognosis were more likely to receive 5-ASA monotherapy than patients with more severe disease. These shifts in the treatment paradigm are a potential explanation as to why preoperative 5-ASA therapy was associated with a decreased odds of infection compared with those not using these agents in studies conducted after 1998. Similarly, it also may explain why there was no increased risk of postoperative infection observed in patients using immunosuppressives for studies conducted after 1998.

In summary, this review found an increased odds of postoperative infectious complications in patients using preoperative corticosteroids and TNF-α antagonists. However, the certainty of this conclusion was considered low, and the authors cautioned that further prospective, well-controlled studies were needed before definitive conclusions could be drawn. The decision to stop therapies before surgery should be individualized to each patient, considering the risks involved including disease flare and sensitization to biologic therapies. Attention should be paid to other modifiable risk factors known to increase the risk of postoperative complications including nutritional status and cigarette smoking.

However, the controversy of whether biologics increase the risk of postoperative infections is yet to be resolved. Since these results were obtained, data from the PUCCINI study, a prospective multicenter cohort study, were published.³⁰ Cohen et al assessed 30-day overall postoperative infectious complications in 955 patients with UC and CD. Patients who received a TNF-α antagonist within 12 weeks of surgery were included. The authors found that TNF-α antagonists were not an independent risk factor for postoperative infections. Further large prospective studies are needed to more definitively understand the risks associated with preoperative medication use in IBD.

RISK OF INFERTILITY AFTER INFLAMMATORY BOWEL DISEASE–RELATED SURGERY

Infertility is a commonly acknowledged complication of IPAA,^{31, 32} which may, in part, be due to scarring of the fallopian tubes during surgery.³³ Previous systematic reviews that have addressed this topic have faced important methodological challenges, including heterogeneity in definitions of infertility, limited capacity to adjust for relevant confounders (ie, age), and relatively limited data.^{31, 32}

A Cochrane review conducted by Lee et al evaluated the effects of IBD-related surgery on female infertility and pregnancy.³⁴ The primary outcome was the risk of infertility defined as the inability to become pregnant after 1 year of regular unprotected intercourse without the use of birth control and infertility after 6 months, 18 months, and 24 months. Secondary outcomes included miscarriage, use of assistive reproductive technology, delivery via caesarean section, stillbirth, preterm birth, low birth weight, and small size for gestational age. A total of 16 studies were included that compared fertility rates in patients with or without a previous surgery (9 studies), open and laparoscopic IPAA (1 study), or before and after surgery (7 studies). A single study compared infertility rates between women with and without IPAA and before and after IPAA. Eight studies included infertility as an outcome; the remainder evaluated pregnancy outcomes.

Two studies including 114 women with UC evaluated the association between previous surgery and infertility after 12 months of unprotected intercourse (RR, 5.45; 95% CI, 0.41–72.57). One study compared women with and without IPAA,³⁵ and the other study compared women with and without restorative proctocolectomy with ileorectal anastomosis.³⁶ A third study evaluated 24-month infertility rates in 86 women with CD and in 104 women with UC. The specific type of surgical procedures were not reported in this study.³⁷ There was a significantly increased risk of infertility at 24 months among women with UC who had surgery (RR, 5.28; 95% CI, 2.9–13.34) but not among women with CD who had surgery (RR, 2.03; 95% CI, 0.56–7.33). A single study compared 12-month infertility rates in 37 women with UC undergoing laparoscopic and open IPAA (RR, 0.70; 95% CI, 0.38–1.27).³⁸

Twelve-month infertility before and after surgery (restorative proctocolectomy with IPAA, restorative proctocolectomy, and total colectomy with ileorectal anastomosis) was described in 5 studies of women with UC. Before surgery, 21% (68 of 327) of women were infertile compared with 63% (239 of 377) after surgery. Similar differences in infertility were noted at 6 months in a single study (before: 20% [1 of 5]; after: 60% [9 of 15])³⁹ and at 24 months in 2 studies (before: 16% [14 of 89]; 71% [116 of 164]).^{39, 40}

In an analysis of infertility before and after surgery stratified by age at surgery, higher rates of infertility were found in patients who underwent surgery after 30 years of age relative to younger patients.⁴¹ This is consistent with age-expected increases in infertility.⁴²

All analyses of the association between previous IBD-related surgery and infertility were rated as very low certainty based on GRADE, given that all studies evaluated were observational, had a high risk of bias, and generated limited data. Thus, important uncertainties exist regarding the association between previous surgery and infertility in women with IBD. The interpretation of these findings is challenging due the heterogeneity in surgical techniques and patient populations. For example, the risk of infertility might be expected to differ between 2- and 3-stage IPAA construction,⁴³ and between hand-sewn and stapled anastomoses.⁴⁴

At present, there is no clear evidence of an association between previous surgery and infertility in women with IBD. This conclusion is based upon the results of the Cochrane review by Lee et al,³⁴ differing from previous reviews on this topic that described a positive relationship.^{31, 32} Unlike previous reviews, Lee et al restricted analyses to studies using a rigorous definition of infertility and did not meta-analyze studies comparing fertility rates before and after surgery. The latter exclusion criterion was used to avoid both the effects of paired data and unmeasured confounding from increasing age in patients after surgery. A more rigorous definition of infertility eliminates bias due to (1) different durations of follow-up between women with and without a previous surgery and (2) the inclusion of women who are voluntarily infertile.

It is important to recognize the heterogeneity in patient populations evaluated in these observational studies. Women with IBD who undergo surgery may be systematically different from those who do not undergo surgery. Specifically, these patients may have more severe disease, long-term disability, or systemic effects of chronic inflammation or corticosteroid use than individuals who do not have surgery. As a result, these individuals may be less likely to want to become pregnant, and their inclusion would result in perceived higher infertility rate. Furthermore, sexual dysfunction rates may be higher among women with prior surgery, which may also contribute to involuntary infertility in this population.^{45, 46}

There is a need for further well-designed studies to evaluate the impact of surgical procedures on infertility in women with IBD. These studies should include detailed information on disease activity and phenotype, medical and surgical treatment, clear definitions of infertility (including the contribution of sexual dysfunction to involuntary infertility), and appropriate comparison groups and statistical analyses that adjust for the effects of important confounding variables.

MEDICAL MANAGEMENT OF SURGICALLY INDUCED REMISSION

Although surgery is an integral part of the management of CD, endoscopic recurrence and clinical relapse rates are high.^47–49 Accordingly, medical therapy for prevention of recurrence after bowel resection is one of the most important unmet medical needs in IBD. Systemic corticosteroids are arguably the most effective induction agents for CD. However, corticosteroids are not effective for maintenance of remission after medical induction therapy.⁵⁰

Although 5-ASAs are highly effective maintenance agents for UC, they are ineffective for maintaining medically induced remission in CD.⁵¹ Likewise, although thiopurines comprised of AZA and 6‐mercaptopurine (6‐MP) are accepted as effective maintenance therapy after medical induction therapy, their role for prevention of postoperative recurrence is controversial, with different systematic reviews reaching different conclusions.⁵¹ This situation has led to international guidelines providing varying advice regarding the efficacy of thiopurines in the postoperative setting. A series of reviews were performed to address these issues, including an update of previous reviews of 5-ASA agents⁵¹ and thiopurines⁵² and a network meta-analysis (NMA) of all medical therapies for preventing postoperative recurrence.⁵³

5-aminosalicylates Agents

Gjuladin-Hellon et al performed a Cochrane review that assessed the efficacy and safety of 5-ASA for preventing postoperative recurrence of CD.⁵¹ The meta-analysis included 14 studies with 1867 participants. The sample sizes of the individual trials ranged from 51 to 324 patients. The risk of bias was considered to be unclear in 7 studies, high in 6 studies, and low in 1 study. Clinical relapse as specified by the included studies was the primary end point.

None of the individual studies showed a statistically significant difference between 5-ASA and placebo. However, a pooled analysis of 5 studies found 5‐ASAs to be significantly superior to placebo for avoiding clinical relapse over a period ranging from 48 weeks to 6 years.^54–58 Relapse was reported in 36% (131 of 361) of 5-ASA participants compared with 43% (160 of 369) of placebo participants (RR, 0.83; 95% CI, 0.72–0.96; moderate certainty evidence). It should be noted that clinical relapse definitions varied across the studies, and only one of the studies in the pooled analysis met a modern FDA definition of symptomatic relapse and endoscopic confirmation of active disease.⁵⁶

Two studies compared sulphasalazine to placebo. Over a period ranging from 18 to 36 months, no differences in the clinical relapse rates were found between the groups. Sixty-six percent (95 of 143) of patients in the sulphasalazine group relapsed compared with 71% (110 of 155) of placebo patients (RR, 0.88; 95% CI, 0.56–1.38; low certainty evidence).^{59, 60} A pooled analysis of 4 studies comparing 5-ASA to AZA showed no significant difference in clinical relapse rates. Sixty-one percent (103 of 170) of 5-ASA patients relapsed compared with 67% (119 of 177) of AZA patients (RR, 0.90; 95% CI, 0.76–1.07; low certainty evidence) at 24 months.^{54, 61–63} In a single trial that compared the TNF-α antagonist adalimumab to 5-ASA, 50% (9 of 18) of 5‐ASA patients relapsed at 2 years compared with 13% (2 of 16) of adalimumab patients (RR, 4.0; 95% CI, 1.01–15.84; low certainty evidence).⁶³

Thiopurines

Gjuladin-Hellon et al assessed the benefits and harms of thiopurines (ie, AZA or 6-MP) for the prevention of postoperative recurrence of CD.⁵¹ Ten studies (928 patients) were included. The risk of bias was thought to be low in 1 study, high in 6 studies, and unlcear in 3 studies. Clinical relapse (defined by the included studies) was the primary end point.

In 3 trials, purine analogues were found with moderate certainty to be more effective than placebo for preventing clinical relapse over a period ranging from 1 to 3 years. Fifty-one percent (109 of 215) of AZA/6‐MP patients and 64% (124 of 193) of placebo patients relapsed respectively (RR, 0.79; 95% CI, 0.67–92).^{54, 64, 65} A pooled analysis of 4 trials found no significant difference in clinical relapse rates between thiopurines and 5-ASA. Sixty-four percent (113 of 177) of thiopurine participants relapsed in comparison with 59% (101 of 170) of 5-ASA participants (RR, 1.05; 95% CI, 0.89–1.24; low certainty evidence) at 24 months.^{54, 61–63}

In a pooled analysis of 3 studies, AZA was found to be significantly inferior to infliximab or adalimumab for prevention of clinical relapse over a period ranging from one to 2 years. Forty-three percent (29 of 67) of AZA participants relapsed in comparison with 14% (10 of 72) of infliximab/adalimumab participants (RR, 2.89; 95% CI, 1.50–5.57; very low certainty evidence).^{63, 66, 67}

Overall, moderate certainty evidence proposes that AZA and 6-MP may be superior to placebo, and very low certainty evidence proposes TNF-α antagonists may be superior to AZA for the prevention of post-surgical relapse.

Network Meta-analysis of Medical Treatments for Maintenance of Remission

The meta-analyses assessing the efficacy of aminosalicylates and thiopurines generated some unanswered questions regarding the relative efficacy of these agents compared with other active medications including TNF-α antagonists. The majority of studies that found a benefit for these therapies were placebo-controlled and relatively small. A GRADE analysis of the results indicates important uncertainty for these conclusions, and the interpretation of how these results can best inform clinical practice is difficult. Adequately powered superiority trials are the best method for comparing the relative efficacy and safety of maintenance therapies in postoperative CD. However in the absence of such trials, an alternative option for obtaining comparative data is an NMA, where different medications are compared using both direct comparisons from RCTs and indirect comparisons across the studies using a common comparator (ie, placebo). That is, if treatment X is compared with treatment Y in one study, and the same treatment Y is compared with treatment Z in another study, indirect information for the comparison of treatment X to treatment Z can be obtained using this technique.

A Cochrane NMA was undertaken to obtain comparative data for interventions in the setting of postoperative CD.⁵¹ Twenty-six RCTs (2581 patients; 9 medications) were eligible for inclusion in the NMA. The 9 medications assessed included adalimumab, 5-ASA, antibiotics, budesonide, infliximab, purine analogues, probiotics, sulfasalazine, and a combination of sulfasalazine and prednisolone. This network lead to 30 direct contrasts, with 102 mixed‐treatment contrasts. The overall quality of evidence was rated as low for both the clinical relapse network (21 studies; 2245 patients) and endoscopic relapse (12 studies; 1128 patients) network due to the imprecision and inconsistency across networks.

Individual contrasts were assessed as very low or low certainty—with the exception 5‐ASA compared with placebo, which was considered to be of moderate certainty. Treatments were ranked based on effectiveness and certainty of the evidence. Results of the NMA are presented in Table 3 and 4.

Clinical relapse was reported in 21 studies in the relapse network (2245 patients). The top 5 ranked treatments included adalimumab, infliximab, budesonide, 5-ASA, and purine analogues. However, limited evidence supports the efficacy of adalimumab for prevention of clinical relapse (hazard ratio [HR], 0.11; 95% credible interval [Crl], 0.02–0.33; low certainty evidence), and moderate certainty evidence supports the efficacy of 5‐ASA for preventing clinical relapse compared with placebo (HR, 0.69; 95% Crl, 0.53–0.87; moderate‐certainty evidence). Budesonide may not be effective for preventing clinical relapse (HR, 0.66; 95% CrI 0.27–1.34; low‐certainty evidence), and lastly, the evidence regarding the effectiveness of infliximab (HR, 0.36; 95% CrI 0.02–1.74; very low‐certainty evidence) and thiopurines (HR, 0.75; 95% Crl, 0.55–1.00; low‐certainty evidence) was uncertain. The certainty of evidence was very low for other medications; therefore, it was unclear if they reduced clinical relapse rates. Nonetheless, further evidence from the PREVENT study (N = 297), which was not included in the NMA due to transitivity issues, suggests that infliximab may be effective for prevention of clinical relapse in high-risk patients.

Endoscopic relapse was reported in 12 studies (1128 patients); however, due to high risk of bias and limited data across the network, the effectiveness of the assessed medications for preventing endoscopic relapse is uncertain.^{55, 63, 65, 66, 68–75} The top 5 ranked treatments for endoscopic relapse included adalimumab, infliximab, antibiotics, purine analogues, and probiotics. Some evidence suggests adalimumab may be effective for prevention of endoscopic relapse (HR, 0.10; 95% Crl, 0.01–0.32; low-certainty evidence); however, none of the other interventions studied appeared to be effective (infliximab HR, 0.24; 95% Crl, 0.01–1.20; antibiotics HR, 0.80; 95% Crl, 0.33–1.65; purine analogues HR, 0.85; 95% Crl, 0.33–1.61; probiotics HR, 1.20; 95% CI, 0.62–2.19). Nonetheless, further evidence from the PREVENT study (N = 297) suggests that infliximab may be effective for prevention of endoscopic relapse.⁷⁶ Further large scale trials are needed to establish the best therapy for prevention of endoscopic relapse in postoperative CD.

Direct analysis has demonstrated that both 5-ASA formulations and thiopurines are probably more effective than placebo for maintenance of surgically induced remission. However, the head-to-head comparisons did not show any difference in efficacy between the medications. Network meta-analyses are significantly impacted by risk of bias, heterogeneity, and imprecision. The analysis demonstrated that only adalimumab and 5-ASA may be effective in maintaining clinical relapse, with no other agent demonstrating efficacy. Future research needs to consider endoscopic relapse and the role of biologics as an outcome of interest, as there is currently insufficient evidence in this area.

CONCLUSIONS

In summary, some general conclusions about the medical management of IBD can be drawn from the symposium presented at DDW 2019.

The efficacy of antibiotics, probiotics, and other interventions for pouchitis are considered uncertain due to limited data and risk of bias. Although broad spectrum antibiotics may be an appropriate first-line therapy for pouchitis, the optimal therapy for both induction of remission and prevention of pouchitis remains unknown. Well-designed, adequately powered RCTs are required to identify the ideal treatment for pouchitis. Preoperative treatment with corticosteroids and TNF-α antagonists may increase the risk of postoperative infectious complications in individuals who have surgery for IBD. However, the certainty of the evidence supporting this conclusion is low due to the observational nature of the data. Further well-designed prospective studies are necessary to draw more definitive conclusions. The decision to stop therapy before surgery needs to be individualized to each patient based on the risk of disease flare and sensitization to biologic therapy.

At present, there is no clear evidence of an association between previous surgery and infertility in women with IBD. Further well-designed studies are required to assess the impact of surgery on infertility in women with IBD. The optimal therapy for the prevention of postoperative recurrence of CD is unknown. 5-ASA may be effective for preventing clinical relapse in some patients. Adalimumab and infliximab may be effective for preventing endoscopic relapse in postoperative CD. Well-designed, adequately powered RCTs are needed to determine the optimal therapy for prevention of clinical and endoscopic relapse in postoperative CD.

Author Contribution: NC, SS, NM, MG, MEK, TMN, JKM, and BGF were presenters and/or involved in the development of the Cochrane DDW symposium. NC, SS, NM, MG, MEK, TMN, JKM, and BGF drafted and/or revised the manuscript for important intellectual content. All authors approved the final draft for submission.

Conflict of Interest: NC has received honoraria for speaking/consulting from AbbVie, Janssen, Takeda, Pfizer, Ferring, Pharmascience, Allergan, Lupin, and Shire.

SS is supported by NIH/NIDDK (K23DK117058), ACG Junior Faculty Development Award, Litwin IBD Pioneers Grant (#623346), and AGA-Pfizer Young Investigator Pilot Research Award in Inflammatory Bowel Disease; he has received research grants from AbbVie and Janssen and personal fees from Pfizer. NN holds a McMaster University Department of Medicine Internal Career Award; he has received honoraria from Janssen, Abbvie, Takeda, Pfizer, Merck, and Ferring. BF has received grant/research support from AbbVie Inc., Amgen Inc., AstraZeneca/MedImmune Ltd., Atlantic Pharmaceuticals Ltd., Boehringer-Ingelheim, Celgene Corporation, Celltech, Genentech Inc/Hoffmann-La Roche Ltd., Gilead Sciences Inc., GlaxoSmithKline (GSK), Janssen Research & Development LLC., Pfizer Inc., Receptos Inc./Celgene International, Sanofi and Santarus Inc., Takeda Development Center Americas Inc., Tillotts Pharma AG, UCB; consulting fees from Abbott/AbbVie, AdMIRx Inc., AgomAB Therapeutics, Akebia Therapeutics, Allakos, Allergan, Amgen, Applied Molecular Transport Inc., Aptevo Therapeutics, Asta Pharma, Astra Zeneca, Atlantic Pharma, Avir Pharma, Biogen Idec, BioMx Israel, Boehringer-Ingelheim, Boston Pharmaceuticals, Bristol-Myers Squibb, Calypso Biotech, Celgene, Elan/Biogen, EnGene, Everest Clinical Research Corp., Ferring Pharma, Roche/Genentech, Galapagos, Galen/Atlantica, GiCare Pharma, Gilead, Gossamer Pharma, GSK, Inception IBD Inc, Intact Therapeutics, JnJ/Janssen, Japan Tobacco Company, Kyowa Kakko Kirin Co Ltd., Lexicon, Lilly, Lycera BioTech, Merck, Mesoblast Pharma, Millennium, Nestles, Nextbiotix, Novonordisk, OM Pharma, Pandion Therapeutics, ParImmune, Parvus Therapeutics Inc., Pfizer, Prometheus Therapeutics and Diagnostics, Progenity, Protagonist, Qu Biologics, Rebiotix, Receptos, Salix Pharma, Shire, Sienna Biologics, Sigmoid Pharma, Sterna Biologicals, Surrozen Inc., Synergy Pharma Inc., Takeda, Teva Pharma, TiGenix, Tillotts, UCB Pharma, Vertex Pharma, Vivelix Pharma, Vifor Pharma, VHsquared Ltd. and Zyngenia; speakers bureau fees from Abbott/AbbVie, JnJ/Janssen, Lilly, Takeda, Tillotts, UCB Pharma; advisory board fees from Abbott/AbbVie, Allergan, Amgen, Astra Zeneca, Atlantic Pharma, Avaxia Biologics Inc., Boehringer-Ingelheim, Bristol-Myers Squibb, Celgene, Centocor Inc., Elan/Biogen, Galapagos, Genentech/Roche, JnJ/Janssen, Merck, Nestles, Novartis, Novonordisk, Pfizer, Prometheus Laboratories, Protagonist, Salix Pharma, Sterna Biologicals, Takeda, Teva, TiGenix, Tillotts Pharma AG, UCB Pharma; and is a senior scientific director at Alimentiv Inc. All other authors have no conficts to declare.

Abbreviations

Abbreviations
 
• 5-ASA

  5‐aminosalicylates

 
• AZA

  azathioprine

 
• CD

  Crohn’s disease

 
• CI

  confidence interval

 
• Crl

  credible interval

 
• GRADE

  Grading of Recommendations Assessment, Development and Evaluation

 
• HR

  hazard ratio

 
• IBD

  inflammatory bowel disease

 
• IPAA

  ileal pouch-anal anastomosis

 
• 6-MP

  6‐mercaptopurine

 
• NMA

  network meta-analysis

 
• OR

  odds ratio

 
• PDAI

  Pouchitis Disease Activity Index

 
• RCT

  randomized control trial

 
• RR

  risk ratio

 
• TNF-α

  tumor necrosis factor-aplha

 
• UC

  ulcerative colitis

REFERENCES