A database literature search for human studies (0–18 years of age) from 1966 to January 2010 on MEDLINE (NLM, National Library of Medicine, Bethesda, MD; 1966–2010), Healthstar on Ovid (1966–2010), and EMBASE (Elsevier, New York, NY; 1980–2010) was performed. The following terms were combined: “ulcerative colitis” with the relevant medical intervention using the MeSH term and all possible text combination (e.g., tacrolimus OR FK506 OR FK-506; ciclosporine OR cyclosporine OR cyclosporin OR ciclosporin; anti-TNF OR infliximab OR remicade). A previously described search strategy was used to determine outcomes of hospitalized patients.10 A search for interventional studies was supplemented by searching the Cochrane Central Register of Controlled Trials (The Cochrane Collaboration, UK) and on the American College of Physicians journal club. References of included articles and review studies were searched for further studies. We included only studies reporting at least five children with acute severe colitis, excluding small case reports. To express the pooled rates across similar strata, studies were weighted according to the inverse variance approach. The pooled analyses were exploratory in nature and are not aimed to elucidate treatment effect of one intervention over the other. Therefore, heterogeneity was addressed clinically and not statistically (i.e., outlying studies were explored for underlying clinical and methodological heterogeneity and subgroup and sensitivity analysis were used).
Incidence of Severe UC in Children
To date, only scarce data exist to evaluate the rate of severe exacerbations in childhood-onset UC. In a retrospective referral cohort, 15/37 children (40%) developed at least one severe acute attack during childhood.11 In a population-based analysis of 196 children with UC under the age of 15 years, 55 (28%; 95% confidence interval [CI]: 23%–34%) required at least one admission for a severe attack during childhood.12 More evidence is available on the prevalence of severe disease specifically at the time of first presentation with UC. In a registry of 97 children with UC under the age of 16 years, 15 (15%) had disease classified as severe at diagnosis.13 In two referral centers, 57% of 171 children newly diagnosed with UC had moderate to severe disease.14
Macroscopic inflammation extending beyond the splenic flexure occurs in up to 80% of children and adolescents with UC, compared to approximately one-third of patients with adult-onset disease.15 This should lead to a higher rate of severe exacerbations in children. An estimated 15% lifetime risk of acute severe UC is reported in the adult literature.16–18 Recent data from Oxford, however, suggest that the risk in adults may be actually similar to children (186 of 750 patients; 25%),19 but referral bias cannot be excluded.
Defining Severity of Disease in Pediatric UC
Several definitions of severe UC are utilized in the adult literature but without appropriate validation.20 The first and most widely used classification of Truelove and Witts21 defines severe disease as the passage of ≥6 bloody stools daily, erythrocyte sedimentation rate (ESR) >30, temperature >37.8°C, pulse rate >90/min, and hemoglobin <10.5 g/dL. The currently accepted modification of the Truelove and Witts' classification mandates the fulfillment of the stool frequency item with at least one of the other four.22 The application of these criteria has never been evaluated in children with severe colitis. In 1977, Werlin and Grand23 proposed a pediatric modification to the Truelove and Witts score defining severe attack as having four out of the following features: >4 bloody stools/day, fever, tachycardia, anemia, and hypoalbuminemia. Both classifications have not been thoroughly validated in children with severe colitis in whom fever is uncommon,24 and pulse rate and hemoglobin are subject to age-related variations. Recently, we developed the Pediatric UC Activity Index (PUCAI) using a prospective cohort of 157 children and a validation cohort of 48 children undergoing colonoscopy2 (Appendix). Cutoff scores for remission, mild, moderate, and severe disease were proposed and validated on an independent “real-life” cohort.25 A PUCAI score of at least 65 points was highly associated with physician global assessment of severe disease (sensitivity 92%, specificity 94% and area under the receiver operator characteristic [ROC] curve 0.99; 95% CI: 0.97–1.0). The PUCAI is noninvasive, requiring only careful inquiry regarding symptoms; the lack of laboratory items enhances feasibility and acceptability for children.12, 25 Unlike the Truelove and Witts classification, which simply determines whether the patient has severe attack or not, the PUCAI yields a range of continuous scores, quantifying the degree of severity, and allowing monitoring of disease progress during treatment. Although the PUCAI performs well in the severe end of the spectrum of UC activity (see below), it nevertheless has the limitation of a “ceiling effect,” meaning that children cannot have scores higher than the maximum of 85 points.
An attempt to standardize criteria for hospitalization and intravenous treatment in acute severe UC is heralded by lack of evidence both in adults and children. Children with very severe disease, those with bowel dilatation, uncontrolled pain, toxic appearance, significant anemia, or hypoalbuminemia should be hospitalized immediately for intravenous corticosteroid treatment. It is important to emphasize that acute severe UC still carries a mortality rate of 1% (mainly from perforation, toxic megacolon, and infectious complications)10; thus, aggressive initial treatment is required. Ominous signs in UC include severe abdominal pain or tenderness, vomiting, fever, significant weight loss, and bowel dilatation.
Initial Management and Feeding
Corticosteroids constitute the first-line therapy of acute severe UC since the landmark trials of Truelove et al.7, 21 Intravenous steroids reduced the mortality rate from 24% to 7%21, 26 in adults. The results from studies evaluating the use of antibiotics in this setting remain controversial and the reader is referred to a previous review on severe pediatric UC.9 There is no firm evidence to support their use and guidelines in adults have not recommended empiric treatment.27 However, it is not unreasonable to initiate empiric antibiotic treatment (such as metronidazole) in children severe enough to require hospitalization, in view of the low quality of studies on the topic, the high frequency of Clostridium difficile infection in admitted patients with UC,28 the relative safety of antibiotics, and the high failure rate of standard medical therapy.
Three comparative studies in adults indicate that bowel rest does not increase remission rate or time to response in acute severe UC.29–31 Hence, fasting is not routinely justified, and is poorly tolerated in children. Nonetheless, food can worsen abdominal cramping and thus if symptoms are severe, a child may benefit from a short period of clear fluids to alleviate symptoms, while bearing in mind that this is not likely to impact treatment outcome. Complete intake restriction is indicated in the uncommon event of surgical abdomen or any other case in which toxic megacolon or perforation is suspected.
Initial evaluation should include blood biochemistry tests including electrolytes, which are often disturbed in severe colitis and are part of the diagnostic criteria for toxic megacolon.32 In addition, complete blood count, albumin, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) should be monitored, with electrolytes, at baseline, and 3 and 5 days thereafter. The absolute value of these tests, and more so the trend of change, indicate response to treatment and may aid in the decision to initiate second-line therapy.10, 33–35 Plain abdominal x-ray is recommended at baseline and thereafter as indicated according to progression of disease activity. Toxic dilatation of the large bowel may be seen with minimal abdominal tenderness and pain. A computed tomography (CT) scan may be of use in highly selected cases of suspected complications and in cases where Crohn's disease (CD) is considered (and in that case, a CT enterography, or even better, magnetic resonance (MR) enterography, is warranted).
Managing abdominal pain with opioids is controversial due to the theoretical risk of precipitating toxic dilatation of the colon.32, 36 This theoretical risk, not supported by solid evidence, should be carefully balanced against the child's pain. It should be remembered that severe abdominal pain, beyond the need for controlling it, is an important marker of disease severity and complications. When no other alternative exists, some use low doses of narcotics (less than 0.1 mg/kg of morphine-equivalent) but this should be given with caution while monitoring for complications; frequent repeated doses should be avoided.
Hypoalbuminemia is a marker for disease severity, has been found in adults to predict response to corticosteroids,10, 33 and may increase surgical complications in children eventually requiring colectomy. However, treatment of hypoalbuminemia is focused on aggressive treatment of the underlying colitis rather than albumin infusion, for which there is no supporting evidence.
Roughly 60%–70% of patients not responding to oral prednisone will respond to the intravenous formulation. Indeed, plasma corticosteroid levels were higher after intravenous injection of prednisolone to six individuals with severe UC compared with standard oral dosing.37 Clinical studies in adults that assessed continuous versus bolus,38 ACTH versus hydrocortisone,8 or pulse dosing,39 failed to identify differences in response. Similarly, in ambulatory patients with moderate disease activity, 40 mg of prednisolone was as effective as 60 mg daily with less toxicity,40 and once daily was as effective as multiple daily doses.41 A heterogeneity-controlled meta-regression in adults found virtually no correlation between steroid dose (at or above 60 mg methylprednisolone daily), and the proportion of patients failing therapy10 (Fig. 1).
Figure 1. Meta-regression of Colectomy rates in 23 cohort studies of hospitalized UC patients treated with various doses of intravenous corticosteroids in the last 30 years, controlled for disease severity at baseline. All doses were standardized as methylprednisolone equivalent. Sum of squares regression line (continuous line) and 95% CI (dotted lines) were calculated after weighting each observation by the inverse variance method (size of the diamonds represents the relative weight of the study on the analysis). Reproduced with permission from Elsevier, UK.10
Download figure to PowerPoint
In children, response to corticosteroids was not associated with dosage within the range administered12, 24 and glucocorticoid bioassay did not predict response to corticosteroids in severe pediatric UC.42 Extrapolating from the adult literature, and based on common clinical practice, it is suggested that the typical pediatric dosing may be 1–1.5 mg/kg daily up to 40–60 mg daily (or hydrocortisone equivalent dose of 200–300 mg daily). The higher doses should be reserved for patients with more severe disease and may be given in two divided doses. Rapid dose reduction to the lower range should be attempted once clinical response is achieved. Further tapering should be slow, over 10–14 weeks.
Abdominal X-ray and Toxic Mega-colon (TMC) in Children
Dilatation of the colon per se is not sufficient for the diagnosis of toxic megacolon, which is better defined as a clinical-radiographic syndrome. Criteria for the diagnosis of TMC in adults were proposed by Jalan et al45 (i.e., the presence of fever, dehydration, hypotension, altered level of consciousness, hematology, and biochemistry blood abnormalities as well as radiographic evidence of a dilated colon). In children some of the criteria are typically absent, including altered level of consciousness and hypotension.32
Studies of acute severe UC in adults report a colon diameter of 65–120 mm in those with TMC versus 10–65 mm in those with severe UC but without the clinical syndrome.46 Abdominal radiographic appearance has been consistently shown to predict response to corticosteroids in adults.4, 5, 33, 47–49 The radiographic criteria for TMC in children are less clear. In the only pediatric study to date carefully examining radiologic features in severe UC, colonic width was not associated with response to therapy.12 Transverse colon diameter of ≥56 mm was strongly suggestive of the clinical syndrome of TMC in children >11 years with IBD.32 In younger children, however, the upper width range was 40 mm.12
Indications for Second-line Therapy
The mortality rate in adults with acute severe UC was reduced from 7% after the introduction of corticosteroids to <1% with timely introduction of second-line therapy.10, 26 Although some case series suggest that prolongation of steroid therapy (i.e., beyond 2 weeks) is effective in eventually achieving remission in some refractory cases,44, 50 this approach is associated with increased toxicity, discomfort, and cost. Delayed surgery for adults with severe colitis not responding to medical therapy was recently found to be associated with postoperative complications.51 Moreover, it is likely that many of the few patients who eventually respond proceed to colectomy within a year of discharge,12, 52 especially those with more active disease during admission.53 Finally, good medical alternatives exist for salvage therapies (see below). Based on these arguments it is thus generally accepted that early introduction of second-line therapy is justified. Clinical guidelines recommend that second-line therapy be commenced if no response to corticosteroids is noted within 3–10 days of initiating intravenous therapy.8, 9, 23, 27, 33, 54, 55
Introduction of second-line therapy should ideally be based on early recognition of patients likely to fail corticosteroid therapy. Four indices in adults have been developed to predict corticosteroid failure on the third treatment day in order to facilitate timely introduction of second-line therapy when needed.33–35, 56 Of these, the Travis score (i.e., the Oxford index) is the most widely used: stool frequency of >8/day or 3–8/day with CRP >45 mg/L on the third day of therapy had a positive predictive value (PPV) of 85% for colectomy.34
In children, a predictive rule based on the PUCAI score at days 3 and 5 of steroid therapy has been implemented12 and proved, in a head-to-head prospective study, to be superior to the four adult scores.24 Aiming at sensitivity on day 3, a PUCAI score of >45 points should dictate planning of second-line therapy (such as infectious screening before antitumor necrosis factor [TNF] therapy and surgical consult). Aiming at specificity on day 5, a PUCAI >65 points, and certainly >70 points, should prompt execution of the planned therapy. This approach was validated on a separate prospective cohort of children with severe UC,57 yielding a PPV of 100% and negative predictive value (NPV) of 87% on day 5. These cutoffs identify roughly 40% of patients requiring second-line therapy, those for whom the likelihood of responding to corticosteroids is less than 8%.12, 57 This is an acceptable false-positive rate considering the potential complications of delaying second-line therapy when required and the option of medical salvage therapy. Half of the others with moderate disease activity on the fifth day (i.e., PUCAI of 35–65) are slow responders and thus should be treated for 2–5 more days with corticosteroids (total 7–10 days of corticosteroid treatment) before a final decision is made. Since the PUCAI is very responsive to change, monitoring the change in score over the next few days could aid in this decision-making.
Choice of Second-line Therapy
Although second-line medical therapy is usually preferred in children, colectomy should always be seriously considered. Colectomy may be life-saving and it ameliorates the need for toxic medication over many future years. However, children and their parents are often less receptive to the idea of ileostomy at a critical time of development when self image is being established. As recently reported among 203 children undergoing restorative surgery, straight ileoanal anastomosis (SIAA) is associated with high stool frequency (mean 8.4 ± 3.9/day at 1 year and 6.2 ± 2.8/day after 2 years), as well a disappointing prevalence of daytime and nighttime incontinence (10% and 29%, respectively, at 2 years). Ileal pouch anal anastomosis (IPAA) is associated with lower rates of daytime and nocturnal incontinence (2% and 20%, respectively, at 2 years), but with the complication of recurrent pouchitis occurring in 49%. Moreover, reduced fertility following IPAA is a major future concern in girls and young women.58–60 Finally, colectomy should be considered with particular care in children younger than 5 years of age, in whom colonic IBD is often relatively difficult to identify confidently as UC versus CD.61
Eight small pediatric cohorts and case series reported the effectiveness of cyclosporine as a salvage therapy in steroid-refractory children with UC with a pooled short-term success rate (i.e., lack of colectomy) of 81% (95% CI: 76%–86%) but only 39% (29%–49%) in the long term (Table 2). The long-term success rate was higher in the two studies that introduced immunomodulators at discharge to all patients (pooled rate 71% (55%–83%)62, 63). Cyclosporine should only be administered as a bridge to thiopurine treatment; its relatively high toxicity profile (Table 2) dictates that it should be discontinued once thiopurines are effective, typically after 3–4 months. In those already on thiopurine before the admission, infliximab should be preferred. In adults, ≈15% of cyclosporine-treated patients experienced significant adverse events including nephrotoxicity, serious infections, seizures, anaphylaxis, and, very rarely, death (but this has also been very rarely reported with infliximab).64
Table 2. Pediatric Studies (All Retrospective Cohorts) of Cyclosporine in Severe Ulcerative Colitis (Only Studies with at Least Five Enrolled Children)
|Study||n||Dose||Response by Discharge||Long-term Colectomy Free||Toxicity|
|Castro 2007 (62)||32||PO 5 mg/kg/d increased to blood level of 150–250 ng/ml (in most cases 7–8 mg/kg/day in 3–4 days)||87%||72% (mean follow up-5 years; all treated with IM)||Severe headaches and parasthesia (n=1), mild headaches (n=4)|
|Socha 2006 (76)||10 (of 23)a||To obtain blood level of 100–200 ng/ml||60%||0% (2months to 5 years)||hirsutism (n=1), gingival hypertrophy (n=1)|
|Barabino 2002 (77)||16 (of 23)a||IV and then PO in varying doses (0.7–7mg/kg/d)||75%||38% (mean follow-up-17 months; IM used in some)||Headache (n=2) seizures (n=1), hypertension (n=2), parasthesia (n=1)|
|Ramakrishna 1996(63)||6 (of 8)a||IV 2 mg/kg/d and then PO, increased to blood level of 100–200 ng/ml||83%||67% (Follow up 2-5 years; all treated with IM)||Tremor (n=2), hirsutism (n=1), all transient|
|Treem 1995 (65)||14||PO 4-8mg/kg/d increased to blood level of 150–300 ng/ml||79%||28% (within 1 year, no concomitant IM)||Mild hypertension (n=5), elevated creatinine (n=4), hirsutism (n=1)|
|Benkov 1994 (78)||5||IV 3 mg/kg/d and then PO, increased to blood level of 400–600 ng/ml||80%||0% (within 1-year, IM started after 6-months)||Headache and significant hypertension (n=1)|
|Treem 1991 (79)||6||PO 4 mg/kg/d increased to blood level of 150–300 ng/ml||83%||33% (with 8 months, no concomitant IM)||Mild hypertension (n=3), elevated creatinine (n=1), mild hirsutism (n=5)|
|Kirschner 1989 (80)||5||IV and then PO, 5-8 mg/kg/d increased to blood level of 100-200 ng/ml||60%||NA||NA|
|Pooled rates (95% CI)||94|| ||81% (76-86%)||39% (29-49%)|| |
Whereas some administer cyclosporine via the oral route from the outset62, 65 (5–6 mg/kg/day, which is equivalent to 2 mg/kg/day intravenously), tacrolimus (FK-506) has better bioavailability and may have a more appealing toxicity profile. Aiming initially for drug levels of 10–15 ng/mL, tacrolimus appears as effective as cyclosporine in adults.66 In children, the data are very limited and show a pooled short-term success rate of 79% (95% CI: 52%–94%) and long-term success rate of only 9% (2%–34%) (Table 3). More data are required to ascertain the role of tacrolimus in acute severe pediatric UC.
Table 3. Pediatric Studies (Both Retrospective Cohorts) of Tacrolimus in Severe Ulcerative Colitis (Only Studies with at Least Five Enrolled Children Were Included)
|Study||n||Dose||Response by Discharge||Long-term Colectomy Free||Complications|
|Lopez 2009 (81)||6 (of 8)||0.12mg/kg to reach levels of 5-10ng/mL||50%||17% (1/6) by 1 year. The last patient underwent colectomy within 6 years||-|
|Ziring 2007 (82)||9 (of 18)a||0.2mg/kg/d to reach levels of 10-15ng/mL initially and then 7-12ng/mL||89%||0% (within a mean of 14 months; tacrolimus used for maintenance instead of IM)||Headache (n=1), tremor (n=3), convulsion (n=1), Lymphoproliferative disease (n=1), mild hypomagnesemia (n=7)|
|Bousvaros 2000 (83)||9 (of 13)a||0.2mg/kg/d to reach levels of 10-15ng/mL||55%||22% (within 1 year, IM started 4 weeks after initiation of tacrolimus)||Headache (n= 2), tremor (n= 3), myalgia (n= 1), hypertension (n= 2), hyperglycemia (n= 2)|
A third alternative for salvage therapy is anti-TNF therapy, which has proved effective in clinical trials in adults with moderate and severe UC.67–69 Six studies of infliximab in refractory severe pediatric UC enrolled at least five children and were included in our analysis (Table 4). The pooled short-term success rate (in most cases meaning discharge without the need for colectomy) was 75% (95% CI: 67%–83%) and the long-term rate was 64% (56%–72%). The data on long-term outcome was heterogeneous with different clinical practices, but infliximab was continued as maintenance therapy in most patients. In an adult study, two or more infliximab infusions were superior to one infusion.70
Table 4. Pediatric Studies on Infliximab in Severe Ulcerative Colitis (Only Studies with at Least Five Enrolled Children with Acute Refractory Colitis Were Included)
|Study||n||Design||Response by Discharge||Long-term Colectomy Free||Complications|
|Turner (43)||33||Prospective cohort||76%||55% (within 1 year, most with maintenance therapy or else IM)||Only one patient discontinued treatment due to infusion reaction|
|Hyams (84)||25 (of 52)a||Prospectively collected registry||68%b||50% (within 1 year, 10/20 with follow-up data)||No opportunistic infections, malignancies or deaths|
|McGinnis 2008 (85)||27 (of 39)a||Retrospective cohort||70%||61% (with a median of 19 months while receiving IM; only some continued infliximab for maintenance)||Infections (n=3), anaphylaxis (n=1)|
|Fanjiang 2007 (86)||27||Retrospective cohort||NA||70% (within a mean of 27 months, while receiving IM and infliximab for 1 year)||Osteomyelitis (n=1), cardiomyopathy (n=1)|
|Russell 2004 (87)||9 (of 14)a||Retrospective cohort||89%||89% (within 6 months; IM was added and infliximab continued as maintenance in some cases)||Mild infusion reaction (n=4)|
|Kugathasan 2002 (88)||5 (of 27)a||Retrospective cohort||80%||NA||NA|
|Pooled rates (95% CI)||126|| ||75%(67-83%)||64%(56-72%)|| |
At present, no randomized comparison of cyclosporine, tacrolimus, and infliximab is available. In a recent prospective nonrandomized study performed in adults,71 52% (23/44) of patients receiving cyclosporine proceeded to colectomy by discharge, but only 18% (5/28) of those administered infliximab (P = 0.003). At 1 year, 68% (30/44) and 44% (11/25) of patients treated with cyclosporine and infliximab, respectively, required surgery for recurrent severe disease (P = 0.049). From the data presented here (Tables 2–4), the short-term success rates in children were remarkably similar with infliximab or calcineurin inhibitors (77%–81%). However, the long-term outcomes differed, with clear superiority for infliximab (64% infliximab versus 39% cyclosporine) (P < 0.001). Keeping in mind the exploratory nature of the analysis, the small sample size, and the heterogeneous methodology, the difference may be explained by the fact that the common practice is to discontinue the calcineurin inhibitor within 3–4 months after discharge while infliximab may be continued as a maintenance therapy. Indeed, a long-term success rate of 71% was achieved by the two studies that commenced azathioprine with cyclosporine. In the thiopurine-naïve patient who responds to infliximab, there are no data to guide choice of continuing infliximab as maintenance therapy versus bridging to thiopurine maintenance. This decision must be made based on local clinical practice and personal preference until more studies are available.
Sequential therapy of cyclosporine and then infliximab or vice versa may be successful in ≈30%–40% of adult patients.72–74 However, based on the current sparse literature, this approach cannot be recommended due to the high associated toxicity and increased surgical morbidity of combined therapy if colectomy is eventually indicated.75