De Novo Crohn’s Disease of the Pouch in Children Undergoing Ileal Pouch-Anal Anastomosis for Ulcerative Colitis

Abstract

Background:

Approximately 10% of children with ulcerative colitis (UC) undergo colectomy with ileal pouch-anal anastomosis (IPAA). We aimed to describe the postoperative outcomes, with an emphasis on chronic pouch inflammation including de novo Crohn disease (CD) at a tertiary care inflammatory bowel disease center.

Methods:

Electronic medical records of all children who underwent colectomy ≤18 years between 2008 and 2017 were reviewed. Clinical and laboratory data were recorded. Primary outcome was frequency of chronic pouch inflammation including de novo CD. Secondary outcomes included early (≤30 days from index surgery) and late postoperative complications. Descriptive statistics (median and interquartile range) summarized the data and univariate analysis tested associations with outcomes.

Results:

Fifty-eight children underwent colectomy and 56 completed IPAA. Median age at diagnosis was 14 years (12–16.2) and at colectomy 16.2 years (14.2–17.7) with median follow-up of 13 months (5–43). Sixty-six percent underwent 3-stage IPAA and 78% were biologic exposed. Eleven had chronic pouchitis, 73% antibiotic refractory and 25% met criteria for de novo CD by median of 19 months (9–41). A total of 21% and 50% experienced early and late surgical complications, most commonly ileus and recurrent IPAA stricture. The pouch failure rate was 3.6%. Chronic pouch inflammation was associated with a later diagnosis of de novo CD (P =.0025).

Conclusions:

In pediatric UC, CD is not uncommon after IPAA. Chronic pouch inflammation often precedes a diagnosis of de novo CD. Families should be informed of the short- and long-term outcomes in children before UC surgery.

Colectomy with ileal pouch-anal anastomosis (IPAA) as a treatment option for pediatric ulcerative colitis (UC) is performed in 9% and 14% at 2 and 5 years from diagnosis, respectively (1). The most common indication for colectomy in children is medically refractory disease (2). Concerns over postoperative complications and long-term pouch outcomes often render surgery a last option.

Inflammation of the pouch, whether acute or chronic, is common in patients post-IPAA. Acute pouchitis can be seen in up to 60% of children after IPAA and chronic pouchitis in up to 33% (3). Chronic pouchitis is often further classified based on the responsiveness to antibiotics. A more concerning inflammatory condition of the pouch that often renders UC surgery as a last resort is the development of de novo Crohn disease (CD), or Crohn-like inflammation. This may be even more amplified in pediatric-onset UC as the published rates have been as high as 28% (4), which is higher than adult IPAA outcome studies (5). The reasons for this are multifactorial and attributable in part to the fact that more children are classified as inflammatory bowel disease unspecified (IBD-U) compared to adults, especially in children with very early onset IBD (<6 years of age) (6). Regardless of age, the inability to differentiate between CD and UC before surgery may be explained by the genetic overlap between CD isolated to the colon and UC (7). In addition to genetics, clinical characteristics and serologic immune markers have been shown to be associated with the development of de novo CD. More specifically, younger age at diagnosis and at time of surgery (8), a family history of CD (9,10), and immunoglobulin A and immunoglobulin G antibodies to Saccharomyces cerevisiae (ASCA) or CBIR1 flagellin (anti-CBir1) (10,11), have been associated with de novo CD.

Herein we describe the postoperative complications and long-term post-IPAA outcomes with an emphasis on de novo CD development in a high surgical volume pediatric IBD referral center.

METHODS

Study Design and Patient Population

As part of an ongoing pediatric IBD postoperative outcomes registry, we conducted a retrospective review of all patients ages 18 and below who underwent colectomy for the diagnosis of treatment refractory UC between January 2008 and December 2017 at the Susan and Leonard Feinstein Inflammatory Bowel Disease Clinical Center, Mount Sinai Hospital, New York, NY. The study was approved by the Institutional Review Board at Mount Sinai Hospital (IRB-18–00140).

Data Collection

Electronic medical records were reviewed for age at diagnosis and surgery, disease behavior (Paris classification), medication exposures, and family history. Additional data collected included laboratory data; complete blood count, complete metabolic panel, erythrocyte sedimentation rate, C-reactive protein, serologic immune markers, histopathology, and any available data on therapeutic drug monitoring. Surgical data of importance included surgery dates, postoperative hospital course, hand-sewn versus stapled anastomosis, and number of stages for IPAA. One-stage surgery was defined as total proctectomy with IPAA construction during 1 procedure. Traditional 2-stage surgery was defined as total proctectomy with immediate IPAA construction and diverting ileostomy with subsequent ileostomy closure. Modified 2-stage surgery was defined as subtotal colectomy (STC) with end ileostomy with subsequent completion proctectomy and IPAA creation. Three-stage surgery was defined as STC with end ileostomy followed by completion proctectomy with IPAA construction and diverting ostomy, and subsequent ostomy closure. Children who underwent IPAA at an outside facility or who had a diagnosis of monogenic IBD were excluded.

Outcomes Measures and Definitions

The primary outcome was the frequency of chronic inflammatory conditions of the pouch, diagnosed clinically based on symptoms and/or endoscopically, and categorized as chronic antibiotic responsive pouchitis: defined as the need for continuous antibiotic use with improvement of symptoms and/or endoscopic/histologic inflammation; chronic antibiotic refractory pouchitis: defined as the need for continuous antibiotic use without improvement of symptoms and/or endoscopic improvement; or de novo CD of the pouch: phenotypes defined based on previously published criteria (12–14): inflammatory CD: the endoscopic presence of either 5 or more aphthae in the afferent limb or deep linear ulcerations in the pouch or afferent limb, stricturing CD: strictures of the afferent limb, pouch inlet, or pouch body, or fistulizing CD: the presence of a fistula developing 3 or more months after ileostomy reversal in the absence of a surgically related local complication, or by the presence of perianal disease at any time.

Secondary outcomes included frequency of early surgical complications, defined as complications within 30 days of index UC surgery (STC); late surgical complications, defined as after 30 days of STC, and/or events in the postoperative period after the second and third stages post IPAA; reoperations; pouch failure, defined as permanent diversion or pouch excision; acute pouchitis, defined clinically based on symptoms and/or endoscopic/histologic pouch inflammation and responding to a short course of antibiotics (≤14 days); and cuffitis, defined as clinical and/or endoscopic inflammation of the rectal cuff.

Statistical Analysis

Standard descriptive statistics, including frequency for categorical variables and median (interquartile range) for continuous variables were calculated unless otherwise stated. Univariate analyses were performed to examine associations of 30-day complications after STC and of pouch outcomes with clinical and laboratory characteristics using Chi square analysis or the Fisher exact test when appropriate. Cumulative incidence of de novo CD of the pouch was determined using R 3.5.1 (The R Foundation for Statistical Computing, 2018). P value <0.05 was considered for significance.

RESULTS

Patient Population

A total of 58 children and adolescents underwent colectomy, with 56 completing IPAA and restoring intestinal continuity. Median follow-up time after ileostomy reversal was 13 months (5–43). Median age at diagnosis was 14 years (12–16.2) and at colectomy 16.2 years (14.2–17.7), with 3 diagnosed below 6 years of age. The median time from diagnosis to colectomy was 19 months (6–35). Using the Paris Classification, 74% had extensive disease (E3 or E4) at the time of diagnosis and 19% were hospitalized at presentation (Table 1). The majority (90%, N = 52) underwent colectomy due to medically refractory disease and the remainder for urgent indications such as toxic megacolon (5%, N = 3) or colonic perforation (5%, N = 3).

TABLE 1.

Patient demographics

	N (%) total 58
Age at diagnosis, y
0–5	3 (5)
6–10	6 (10)
11–18	49 (85)
Male sex	37 (66%)
Disease location at diagnosis (Paris classification); N = 38
E1	2 (5.3)
E2	8 (21)
E3	7 (18.4)
E4	21 (55.3)
First-degree relatives with IBD	14 (24)
Second-degree relatives with IBD	11 (19)
Precolectomy IBD serologies; N = 21
Anti-Saccharomyces cerevisiae antibody positivity	4 (19)
Perinuclear neutrophil antibody (pANCA)	18 (86)
Median (IQR) BMI z score at first surgery	−1.11 (−2.1 to −0.1]
Median (IQR) hemoglobin at first surgery, g/dL	10.3 (9.1–12.3)
Median (IQR) platelets at first surgery, × 103/μL	397.5 (305.3–505)
Median (IQR) albumin at first surgery, g/dL	3.3 (2.7–3.8)
Median (IQR) ESR at first surgery (0 −15 mm/h)	28.5 (12.8–71.3)
Preoperative biologic exposure	45 (78)
Anti-TNF only	32 (71.1)
Vedolizumab only	1 (2.2)
Anti-TNF and vedolizumab	11 (24.4)
Anti-TNF and vedolizumab and ustekinumab	1 (2.2)
Median (IQR) age at colectomy, y	16.2 (14.2–17.7)
Median (IQR) time from diagnosis to colectomy, mo	19 (6–35)
Median (IQR) time from diagnosis to first biologic, mo	3.5 (1.2–13.7)
Median (IQR) time from biologic to surgery, mo	4.2 (1.8–14.6)
Laparoscopic colectomy	46 (79)
Colon pathology
Transmural inflammation	2 (3.4)
Epithelioid granulomas	0 (0)
Cytomegalovirus colitis	1 (1.7)
IPAA number of stages, N = 56
1-Stage	3 (5)
2-Stage	16 (29)
3-Stage	37 (66)
Hand-sewn IPAA anastomosis	21 (36)
Median (IQR) time to last follow-up, mo	13 (5–43)

BMI = body mass index; ESR = erythrocyte sedimentation rate; IBD = inflammatory bowel disease; IQR = interquartile range; TNF = tumor necrosis factor; IPAA = ileal pouch-anal anastomosis

A laparoscopic approach for colectomy was feasible in 46 (79%) and the remainder were performed by an open approach. From 2013 onward, there were significantly more laparoscopic approaches performed compared with open colectomy (P < 0.00001). Of the 56 who underwent IPAA, 37 underwent a 3-stage (66%), with the remainder either 2-stage (29%) or 1-stage (5%) IPAA. The rate of 3-stage IPAA increased from 48% between 2007 and 2015 to 88% between 2016 and present (P =.0021) (Table 2). A hand-sewn anastomosis was performed in 36% of IPAAs.

TABLE 2.

Number of ileal pouch-anal anastomosis stages and colectomy approach by calendar year

Year	Colectomies N	Laparoscopic colectomy N (%)	Open colectomy N (%)	IPAA N	1 or 2 Stage IPAA; N (%)	3 Stage IPAA; N (%)
2007	1	0 (0)	1 (100)	N/A	N/A	N/A
2008	7	2 (29)	5 (71)	5	4 (80)	1 (20)
2009	0	0 (0)	0 (0)	3	1 (33)	2 (67)
2010	8	5 (62.5)	3 (37.5)	6	4 (67)	2 (33)
2011	6	4 (67)	2 (33)	5	2 (40)	3 (60)
2012	3	2 (67)	1 (33)	5	3 (60)	2 (40)
2013	4	4 (100)	0 (0)	3	1 (33)	2 (67)
2014	3	3 (100)	0 (0)	2	1 (50)	1 (50)
2015	7	7 (100)	0 (0)	2	0 (0)	2 (100)
2016	11	11 (100)	0 (0)	15	3 (20)	12 (80)
2017	8	8 (100)	0 (0)	7	0 (0)	7 (100)
2018	N/A	N/A	N/A	3	0 (0)	3 (3)
Total	N = 58	N = 46 (79%)	N = 12 (21%)	N = 56	N = 19 (34%)	N = 37 (66%)

IPAA = ileal pouch-anal anastomosis.

With regards to steroid exposure, 50% were exposed to at least 20 mg/day within 1 month of index surgery. Seventy-eight percent (N = 45) were biologic-exposed before colectomy, the majority with anti-tumor necrosis factor (TNF) therapy alone (71%), followed by both anti-TNF and vedolizumab (24%), vedolizumab only (2%), and anti-TNF, vedolizumab and ustekinumab (2%). Of these, 60% received a biologic within 1 and 3 months before index surgery, and 1/3 received a biologic within 6 months before index surgery. In 6 children, the first biologic exposure occurred within 1 month before colectomy. Of those who received a biologic within 1 month of index surgery, 78% received anti-TNF therapy and 22% vedolizumab. Of those exposed to infliximab at time of colectomy, 8 children had measured infliximab drug concentrations, with 4 of these above 34 μg/mL, and the others at 2.9, 6.6, 8.6, and 24 μg/mL.

Before index surgery, nearly all children were found to be anemic, with median hemoglobin of 10.3 g/dL (9–12.3). More than half (53%) had hypoalbuminemia, with median albumin 3.3 g/dL (2.7–3.8), and elevated C-reactive protein in 72%. Twenty-one (36%) had IBD serologies pre-colectomy. Perinuclear antineutrophil cytoplasmic antibody (pANCA) was positive in 18 (86%) and 10 had a level of 100 EU/mL and above. ASCA was positive in 4 (19%) and Cbir-1 in 2.

Inflammatory Pouch Outcomes

Acute Pouchitis

Episodes of acute pouchitis occurred in 50% (N = 28) with 11 children (20%) having recurrent episodes (median 1 [1–2]). Eleven of the 28 (39%) developed antibiotic-dependent chronic pouchitis. Three developed de novo CD without progressing to chronic pouchitis. Episodes were most commonly treated with ciprofloxacin and metronidazole and less commonly with amoxicillin/clavulanic acid, rifaximin, levofloxacin and trimethoprim. Median time from ostomy takedown to first episode of acute pouchitis was 5 months (1–11).

Cuffitis

Nine of the 56 (16%) experienced 1 or more episode of cuffitis. In one of these children the diagnosis of inflammatory CD was already made and cuff inflammation responded to adalimumab and azathioprine. Another was lost to follow-up. In the remaining children, 43% received rectal therapy only, 43% a combination of rectal therapy and antibiotics (ciprofloxacin, metronidazole, or trimethoprim/sulfamethoxazole), and 14% antibiotics only (metronidazole). Eight of the 9 with cuffitis had a stapled anastomosis; the remaining 1 patient had a hand sewn anastomosis however end up having a 5 cm residual rectal mucosa left behind.

Chronic Pouchitis

Eleven children had chronic pouch inflammation (20%) with 8 being antibiotic responsive initially and 5 of those progressing to antibiotic refractory. The remaining 3 were classified as antibiotic refractory only. Ten of the 11 (91%) were boys. Two children with antibiotic responsive chronic pouchitis with precolectomy IBD serologies had a pANCA level >100 EU/mL. Six with chronic antibiotic refractory pouchitis eventually met criteria for de novo CD of the pouch (Fig. 1A).

FIGURE 1.

A, Overview of pouch inflammation and progression to de novo Crohn’s disease (CD) of the pouch. B, Cumulative incidence of de novo Crohn’s Disease of the pouch. Months indicates time from ileostomy closure. *Five additional children developed de novo CD without any prior pouch inflammation noted.

De Novo Crohn Disease of the Pouch

Fourteen of the 56 who underwent IPAA (25%) met the criteria for de novo CD by a median of 19 (12–46) months after ileostomy closure (Fig. 1B). The median age at UC diagnosis was 14.2 years (12.3–16.7), which was similar to those who were not diagnosed with de novo CD as of last follow-up, and the median age at diagnosis of de novo CD was 18.4 years (17–20.5). Six of 14 were categorized as inflammatory phenotype (43%), 4 fistulizing (29%), 3 fibrostenotic (21%), and 1 met criteria for both fibrostenotic and fistulizing (7%). All fistulas but 1 (pouch-labial) was perianal.

At last follow-up, 6 of 14 (43%) were on treatment with biologics: 3 with anti-TNF and 3 with vedolizumab. In 2, antibiotics were used in combination with vedolizumab and 3 were on antibiotics alone at last follow-up. Two children had pouch excision with permanent ileostomies: 1 in the face of fibrostenotic disease despite anti-TNF therapy after 50.7 months, and the other with severe perianal fistulas requiring multiple diversions and anti-TNF therapy after 85.3 months. One additional child with inflammatory CD had a diverting loop ileostomy at last follow-up.

Based on univariate testing, there were no clinical or serologic marker that was associated with de novo CD including ASCA or CBir1 status, sex, younger age, or family history of IBD. There was, however, a positive association between a history of any chronic pouch inflammation and a diagnosis of de novo CD of the pouch (P = 0.0025).

Postoperative Complications

Early Postoperative Complications

Within 30 days of the index UC surgery, 12 of 58 (21%) experienced a total of 15 complications. The most common complication was postoperative ileus (N = 7), followed by small bowel perforation (N = 2), intra-abdominal abscess (N = 2), portal vein thrombosis (N = 2), peristomal abscess (N = 1), and small bowel obstruction (N = 1). Four children underwent a total of 5 reoperations as a result of the early complications. There were no clinical, laboratory, or treatment exposure factors that were associated with 30-day complications (Table 3).

TABLE 3.

Risk factors for complications within 30 days of after index surgery

	Complication (n)
Risk Factor	Yes	No	P
Male sex	7	31	0.56
Extensive colitis (E3, E4)	5	23	0.29
Open (nonlaparoscopic) surgery	3	9	0.70
1- or 2-Stage surgery	6	12	0.14
Anemia (Hgb ≤10 g/dL)	2	19	0.20
Thrombocytosis (≥300 × 103/μL)	7	33	0.31
Elevated CRP	5	0	0.29
Vitamin D deficiency (≤30 ng/mL)	0	7	0.46
BMI z score ≥1	0	5	0.57
Steroid exposure preop
Within 30 days	5	24	0.28
90 Days	2	13	0.45
180 Days	1	8	0.66
Any biologic exposure preop
Within 30 days	4	23	0.35
90 Days	5	22	0.63
180 Days	4	12	0.71
Anti-TNF exposure preop
Within 30 days	3	18	0.49
90 Days	1	15	0.08
180 Days	0	9	0.17

BMI = body mass index; CRP = C-reactive protein; Hgb = hemoglobin; TNF = tumor necrosis factor.

Late Postoperative Complications

A total of 51 late postoperative complications occurred >30 days after colectomy or any time following IPAA in 28 of 56 (50%) children resulting in 40 reoperations. The most common were ileus (N = 15) and recurrent anastomotic stricture (N = 11), followed by small bowel obstruction (N = 9), anastomotic leak (N = 5), wound thrombosis (N = 3), intra-abdominal infection (N = 4), portal vein infection (N = 3), and skin dehiscence (N = 1). Five of the 51 occurred after colectomy and before IPAA and included ileus in 3 and small bowel obstruction in 2. The remaining 46 complications occurred after J-pouch construction. Median time to complication was 1.1 months (0.4–6). Long-term complications occurred in 67% of 1-stage, 56% of 2-stage, and 46% of 3-stage surgeries. Other complications included pouch prolapse (N = 3 among 2 children), pouch excision (N = 2), and pouch dehiscence (N = 3). The total pouch failure rate was 3.6% (N = 2).

DISCUSSION

As part of our ongoing postoperative pediatric IBD registry, we have demonstrated that a diagnosis of de novo CD after IPAA for treatment refractory childhood onset UC is not uncommon, occurring in one quarter of the population. Our data also suggest that children given a diagnosis of de novo CD often progress from 1 of 2 chronic pouch inflammatory states with antibiotic refractory being the most common preceding diagnosis. This was also supported by Alexander et al (15), who described a 91% conversion rate of those with chronic pouchitis who later developed CD.

CD of the pouch after IPAA has been reported in pediatrics with a broad range of incidence, likely related in part to the discrepancy of follow-up times and criterion used. Mortellaro et al (16) demonstrated a change in diagnosis to CD after IPAA in 13% after a mean of 2 years (range 1–5) which was similar to our follow-up time of median 19 months. To define de novo CD, our definitions encompassed those previously published by large volume IPAA centers (12–14). It should be noted that there is no clear consensus on whether pure inflammation of the pouch can be categorized as de novo CD as opposed to chronic pouch inflammation. This, however, may purely be semantics as inflammation that is chronic and refractory to antibiotics should raise the question of de novo CD and managed accordingly. The use of biologics in the case of antibiotic refractory pouchitis has been published of late raising the question as to whether the biology of chronic pouch inflammation more so resembles that of IBD and not pouchitis (17,18). More research is needed to understand this phenotype; however, the large genetic overlap between CD and UC for the identified susceptibility loci may render the true distinction nearly impossible. In pediatrics this may even be amplified by the isolated colonic phenotype we more commonly see in the younger age group and the higher likelihood of carrying a diagnosis of IBD-U.

An attempt to predict which children with UC are at highest risk of de novo CD has been the focus of many investigators. The goal of this work is less to deter a sick, treatment refractory patient from undergoing an STC, but to help guide postoperative management and possibly prevent a devastating outcome like pouch failure and pouch excision (19,20). For example, in a patient who has ASCA positivity precolectomy and a family history of CD, Melmed et al (21) showed that this patient is at high likelihood of developing a more typical CD phenotype of the pouch. It is conceivable that such a patient would not undergo an IPAA and perhaps an ileorectal, or if electing for an IPAA, receive postoperative prophylaxis akin to what is commonly done post-ileocecal resection. Other serological markers such as pANCA level above 100 EU/mL in particular, have been shown to be associated with chronic pouch inflammation (22). In this circumstance the use of prophylactic antibiotics may be the correct strategy. This, however, has never been studied prospectively. Multiple other factors alone or in combination have been explored and a predictive model has not been developed to date that predicts with absolute or high certainty that a patient will develop a CD phenotype once the fecal stream has been restored (8,12,23).

Beyond pouch inflammation, pediatric outcomes in the immediate and later postoperative period associated with the surgery itself merit further description. Koivusalo et al (24) reported similar findings to ours with rates of 38% early and 40% late complications. Portal venous thrombosis is of special interest and is usually diagnosed when a patient comes to medical attention for presumed bowel complication and it is serendipitously found on imaging. Our rate was 8.6% and previous studies report an incidence of up to 18% in children after colectomy even despite a 64% rate of thromboprophylaxis (25). This is likely an underestimation as most centers are not routinely screening for thrombi (26) and because many are asymptomatic (27). Providers should have a high suspicion for portal vein thrombi when children present postcolectomy with abdominal pain, nausea, or vomiting.

Adult guidelines support the use of anticoagulant thromboprophylaxis during hospitalizations in those with IBD (28) and for 4 weeks after IBD surgery (29). At our center in recent years we have implemented routine use of medical thromboprophylaxis with either subcutaneous heparin or enoxaparin sodium postoperatively in all pediatric patients with UC during the hospital stay as the postoperative risk for thrombosis is high, especially in the face of severe, active colitis (30). The need for postdischarge continuation of these therapies varies based on risk factors and is assessed on a case-by-case basis.

This study was limited by the retrospective design and small number of patients who underwent IPAA and followed at our center. The small cohort size is not uncommon in pediatric UC especially with the increased utilization of biologics and the expanding number of targets. Given this is an ongoing registry we will be able to track outcomes even further to have longer follow-up times, which may influence our outcomes.

Our data suggest that children undergoing IPAA for presumed UC carry a risk as high as 25% of being diagnosed with de novo CD especially if they have been classified as having chronic pouchitis. Efforts are underway to focus more on the biology of the pouch perhaps at a molecular level so that predictive models can be developed either pre- or post-IPAA to proactively treat patients at highest risk of developing de novo CD. Children and families should be informed about the risk of de novo CD especially in children diagnosed with IBD-U so more informed decisions can be made as to the postoperative management strategy.

What Is Known.

• Complications are common after colectomy and ileal pouch-anal anastomosis.

• In pediatric-onset inflammatory bowel disease it is challenging to definitively diagnose Crohn disease versus ulcerative colitis.

• A subset of these patients later develop Crohn disease after ileal pouch-anal anastomosis and develop related pouch complications.

What Is New.

• Chronic pouch inflammation is associated with de novo Crohn disease after ileal pouch-anal anastomosis.

• A change in diagnosis from ulcerative colitis to Crohn disease is made in 25% of pediatric-onset inflammatory bowel disease.