Kelley-Quon, Lorraine I.*; Jen, Howard C.*; Ziring, David A.*; Gupta, Neera‡; Kirschner, Barbara S.§; Ferry, George D.||; Cohen, Stanley A.¶; Winter, Harland S.#; Heyman, Melvin B.‡; Gold, Benjamin D.**; Shew, Stephen B.*
*Mattel Children's Hospital
‡UCSF Benioff Children's Hospital, University of California, San Francisco, San Francisco, CA
§The University of Chicago Comer Children's Hospital, Chicago, IL
||Texas Children's Hospital, Baylor College of Medicine, Houston, TX
¶Children's Center for Digestive Healthcare, Children's Healthcare of Atlanta, Atlanta, GA
#MassGeneral Hospital for Children, Boston, MA
**Emory University School of Medicine, Atlanta, GA.
Address correspondence and reprint requests to Stephen B. Shew, MD, Division of Pediatric Surgery, Mattel Children's Hospital, David Geffen School of Medicine at UCLA, 10833 Le Conte Avenue, CHS Bldg, MC 957098, Los Angeles, CA 90095-7098 (e-mail: firstname.lastname@example.org).
Received 9 February, 2012
Accepted 25 May, 2012
L.K.Q. is a 2010–2012 Robert Wood Johnson Foundation Community Health Leader. M.H. is supported in part by NIH grant DK060617. N.G. is supported in part by NIH grant DK077734. B.K. is supported in part by grants from Abbott and Janssen Biotech. S.C. is supported in part by grants from Given Imaging, Janssen, Abbott, UCB, and Astra Zeneca. H.W. is supported in part by grants from Janssen, UCB Pharma, Nutricia, and Prometheus Labs. B.G. is supported in part by grants from Takeda Pharmaceuticals North America. S.S. is supported in part by NIH grant HD052885. The other authors report no conflicts of interest.
The natural course of UC in children is often more severe than that in their adult counterparts (1). Twenty to 30% percent of all cases of inflammatory bowel disease (IBD) are diagnosed during childhood (2), with ulcerative colitis (UC) affecting 22% of all of the children diagnosed with IBD (1). For the pediatric patient, UC is a lifelong disease making optimization of medical and surgical therapy paramount. Although significant advances have been made in medical management of UC, a large proportion of children ultimately require restorative proctocolectomy for definitive control of their disease.
Early studies of UC in children attempted to define severity of illness and risk of colectomy based on the number of bloody stools per day. This measurement can be extremely subjective and difficult to replicate. Although more reports have found stooling patterns to be greatly predictive of failure of corticosteroid management (3), these reports relate to hospitalized acute severe colitis with short-term outcome evaluation, which differs from the majority of children with UC who are ambulatory and receive care over a longer duration. It is possible that a more accurate definition of disease severity would take into account patient history and laboratory findings in addition to stooling patterns. Preliminary studies in France and Japan have identified such clinical predictors of colectomy in adults with UC, but few studies have been performed in the United States, and fewer in children (1,4).
Unfortunately, children may present to the surgeon after a prolonged course of failed medical management and immunosuppression, with surgery rates as high as 60% in children with steroid refractory UC (5). Patients with a prolonged severe course of medically refractory disease often require prolonged hospitalization and experience increased postoperative complications (6). Furthermore, although reports have explored indicators for surgery in children with Crohn disease (7), few studies have identified predictors of surgery in children with UC. The primary goal of the present study was to use a large, multicenter disease-specific database to identify clinical predictors of surgical intervention in children with UC.
This retrospective cohort study was performed after obtaining institutional review board approvals from each individual institution and the Pediatric Inflammatory Bowel Disease Consortium (PediIBDC). Data were obtained from the PediIBDC Database, a registry established in January 2000 that collects demographic, clinical, and epidemiologic data reported by investigators at hospitals specializing in pediatric IBD care. The data used for our study reflected information submitted from 7 regional IBD centers, including the present authors’ associations and the Barbara Bush Children's Hospital at Maine Medical Center (8). Subjects eligible for inclusion in the database include all of the patients with IBD (UC, Crohn disease, and inflammatory bowel disease-unspecified [IBD-U]) diagnosed before 18 years of age and actively followed in a pediatric gastroenterology clinic at 1 of the participating hospitals. The initial and final diagnoses of IBD are recorded in the database. Children with existing and newly diagnosed IBD are prospectively enrolled. Data for patients examined in this report were retrieved from January 1999 through November 2003. This time period represents the original and most complete version of the PediIBDC dataset. Eligible patients included all of the children enrolled in the database with a final diagnosis of UC at the time of enrollment through November 2003. UC is defined in the database as continuous disease confined to the colon and no evidence of small intestinal disease (other than backwash ileitis) on biopsy or radiographically (8). The diagnosis of UC was present for each patient at the initiation of and through the entire course of clinical care. For example, if a patient was diagnosed with Crohn disease discovered on surgical pathology after colectomy, they were not included in the cohort.
Predictor and Outcome Variables
Demographic information including age at diagnosis, sex, race, body mass index, and time interval before surgery was extracted from the database. Clinical history including family history of IBD (including Crohn disease, UC, and IBD-U) and preoperative medical therapy (5-ASA [5-aminosalicylate], 6-MP [mercaptopurine], azathioprine, azulfidine, cyclosporine, infliximab, methotrexate, steroids, and tacrolimus) was also retrieved. Symptoms present at time of diagnosis of IBD including abdominal pain, abnormal laboratory reports, diarrhea, joint pain, fatigue, fever, nausea, poor growth, dermatologic complications, rectal bleeding, vomiting, and weight loss were also included in our analysis. Initial laboratory values measured at the time of diagnosis including serum erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), albumin, and hemoglobin were also evaluated. Although many of the laboratory values evaluated vary with age and sex, clinically relevant laboratory cutoffs were defined (ESR > 20 mm/h, CRP > 5.0 mg/dL, and hemoglobin < 10 g/dL) as no significant association was appreciated on bivariate analysis of continuous laboratory values. Albumin was initially analyzed in the regression as a continuous variable but a cutoff of <3.5 g/dL was later found to be statistically significant (see discussion below).
All potential predictors were evaluated at time of diagnosis. For children prospectively enrolled (those with new-onset disease from January 1999 to November 2003), time of diagnosis coincided with time of enrollment. For children retrospectively enrolled (those with existing disease diagnosed before November 1999), a retrospective chart review identifying the clinical encounter during which time they were diagnosed with UC was used to identify symptoms and laboratory values present at time of diagnosis. There were no missing data for any parameter in this cohort.
Bivariate statistical comparisons were made using χ2 and Student T-statistics stratified by surgery. Factors associated with colectomy 2 years after diagnosis and overall colectomy were identified using Cox proportional hazards regression analysis. Patient demographics, clinical history, presenting symptoms, and laboratory values were included in the models as covariates. Statistical analyses were performed using SAS version 9 (SAS Inc., Cary, NC) and Stata/IC version 10.1 (Stata Corp, College Station, TX). Because many variables were analyzed as potential surgical predictors, a P-value of <0.01 was considered significant to counterbalance the risk of a Type I error owing to multiple comparisons.
Query of the PediIBDC Database yielded 406 children with UC eligible for inclusion (Table 1). Approximately half of the cohort was boys (49%) and white (42%) with an average age at diagnosis of 10.6 ± 0.22 years. Of the 57 (14%) children who underwent colectomy, median time to surgery was 3.8 (interquartile range 4.9) years after diagnosis (Fig. 1). Average follow-up was 6.8 (±4.0) years. One- and 2-year colectomy rates were 4% (n = 18) and 7% (n = 30), respectively. Bivariate statistical comparisons of patient demographics, presenting symptoms, and laboratory values stratified by surgery revealed there was a higher proportion of Hispanic (P = 0.04) and non-white children (P = 0.042) in the surgical cohort, but these children represented a small minority of the overall population of the present study. Additionally, there was a higher proportion of children presenting with weight loss at the time of initial diagnosis in the surgical cohort compared with the nonsurgical cohort (P < 0.006). Comparisons of family history and use of immunosuppressants stratified by surgery revealed that the use of cyclosporine or tacrolimus (P < 0.001 and P < 0.001) at any time during medical treatment was also seen in a higher proportion in the surgical cohort.
Statistically significant clinical factors identified in bivariate analysis (weight loss at diagnosis, medical therapy) and factors thought to be clinically relevant (age, sex, white race, laboratory values present at time of diagnosis, family history of IBD) were included in the multivariate analysis (Table 2). Children presenting with weight loss (hazard ratio [HR] 2.55, P < 0.001) or serum albumin < 3.5 g/dL (HR 6.05, P < 0.001) at the time of diagnosis, and children with a first-degree relative with UC (HR 1.8, P < 0.001) required earlier surgical intervention than children without these characteristics. Furthermore, children treated with cyclosporine (HR 6.1, P < 0.001) or tacrolimus (HR 3.66, P < 0.001) also required earlier surgical management of their UC compared with children not treated with calcineurin inhibitors. Of note, albumin <3.5 g/dL at time of diagnosis and having a first-degree relative with UC were not predictive of 2-year risk of colectomy, whereas they were predictive of overall risk of colectomy. History of azathioprine (HR 0.06, 99% confidence interval [CI] 0.01–0.27) or 6-MP (HR 0.21, 99% CI 0.05–0.88) use was associated with a decreased 2-year risk of colectomy, whereas these factors did not affect overall risk of colectomy. Each factor predictive of overall colectomy was associated with an increased rate of 2-year and overall colectomy (Table 3); however, for albumin <3.5 g/dL and having a first-degree relative with UC, the risk of colectomy 2 years after diagnosis was notably attenuated (Fig. 2).
This is the first study to use a large, multicenter disease-specific database to identify clinical predictors of colectomy in children with UC. Our analysis identified hypoalbuminemia and significant weight loss at time of diagnosis as predictive of earlier colectomy, indicating that nutritional status and/or extent of protein losing enteropathy play an integral role in the clinical severity of UC in children. This finding underscores the importance of thoroughly assessing nutritional status at the time of diagnosis of UC. We also found that having a first-degree relative with UC and previous use of cyclosporine and tacrolimus predicted an earlier need for surgery. Overall, our results indicate that children with more severe disease burden at time of diagnosis likely do not have sufficient physical reserve (ie, low albumin, immunosuppression, muscle wasting), develop recalcitrant symptomatology and ultimately require colectomy for optimal disease control.
Poor nutritional status has previously been identified in the literature as predictive of clinical course in patients with UC. Low albumin levels after first induction therapy were reported by Shiga et al (4) to predict UC relapse and need for colectomy; however, the study of Shiga et al did not find low albumin at time of diagnosis to be predictive of surgery. In our initial analysis, we examined albumin as a continuous variable and found that it was not independently predictive of colectomy. When albumin was analyzed, however, as a dichotomous variable with a cutoff of <3.5 g/dL, albumin levels became significantly predictive in our model. This implies that once a nutritional threshold is passed, children with UC are likely not able to recover as robustly as their more healthy counterparts. Furthermore, >20% of our subjects had an albumin <3.5 g/dL, indicating that our cohort was likely to detect a difference between groups.
The greatest risk factor for developing UC in childhood is having a first-degree relative with UC, even though lesser than the genetic contribution observed in Crohn disease (9,10). Our results agree with reports that there is a hereditary component to UC, although the UC phenotype is likely more driven by environmental factors. The present findings imply that children with a first-degree relative with UC are likely to have more severe clinical manifestations of UC, thereby requiring colectomy at an earlier date. Although multiple genetic contributors of UC have been established (11), the pathogenesis of UC is likely multifactorial and those with a familial predisposition may express a more severe phenotype than those without a genetic contribution.
Children with a more severe disease burden typically require more potent immunosuppressive regimens and require colectomy at a higher rate. Tacrolimus and cyclosporine are used most commonly in steroid-refractory UC, so it was not surprising that use of these calcineurin inhibitors was greatly predictive of early colectomy in the cohort. Our results imply that calcineurin inhibitors are likely a proxy for disease severity. This is in agreement with prior studies reporting colectomy rates in up to 60% of children with steroid refractory UC treated with tacrolimus (5). Furthermore, our results also coincide with studies showing that the use of azathioprine, 6-MP, and infliximab does not predict colectomy (12). Of note, 1 limitation of our study is that the use of infliximab for refractory UC has increased in the last 10 years and the present study took place before this clinical trend was more established. Our findings, however, disagree with the report of Tremaine et al of increased colectomy in patients receiving aminosalicylate, prednisone, and azathioprine/6-MP therapy (13). The final regression analysis in the latter study did not examine the use of more potent immunosuppressants such as cyclosporine and tacrolimus as the number of participants using these drugs was small. Owing to the exclusion of calcineurin inhibitors, it may be that maximal medical therapy was thereby defined as patients who received aminosalicylate, prednisone, azathioprine, or 6-MP. Patients receiving these drugs therefore had a more severe disease burden compared with the rest of their cohort and therefore were in higher proportion in the colectomy arm. Additionally, Tremaine et al. conducted a single-institution study in a predominantly adult population, and therefore the external validity of the conclusions of the study may be limited in the pediatric setting.
Factors predictive of 2-year risk of colectomy did not entirely coincide with those predictive of overall risk of colectomy. Albumin <3.5 g/dL at time of diagnosis and having a first-degree relative with UC were not predictive of 2-year risk of colectomy, whereas they were predictive of overall risk of colectomy. This implies that these factors may have less effect on short-term risk of colectomy for children with UC but maintain significance when predicting overall risk of colectomy. Furthermore, history of use of azathioprine or 6-MP was associated with a decreased 2-year risk of colectomy, whereas these factors did not affect overall risk of colectomy. This latter finding likely implies that the use of azathioprine and 6-MP may delay early disease progression significantly for children with UC but that overall, their long-term clinical benefit may be uncertain. Finally, weight loss at diagnosis and previous use of cyclosporine and tacrolimus remained significant predictors of both 2-year and overall colectomy risk. This finding underscores the prognostic significance of a child's nutritional status at the time of diagnosis and also affirms the use of calcineurin inhibitors in the present study cohort as the probable limit of medical therapy for UC before proceeding with colectomy.
Contrary to the most recent single-center report by Moore et al (14), low hemoglobin and leukocytosis at the time of diagnosis did not predict colectomy in our analysis. This may be because of the abnormally high colectomy rate reported in Moore cohort (16.7% at 1 year and 35.6% at 3 years). The overall cumulative rate of colectomy in children with UC is typically lower, with rates reported from 5% to 8% at 1 year after diagnosis, and up to 20% at 5 years (1,15). The overall colectomy rate in our study was 14% with a median time to surgery of 3.8 years, which is closer to reported average colectomy rates for children with UC. Owing to the profound effect UC has on overall development and growth in children, we agree with the authors in concluding that creation of a risk stratification tool would facilitate counseling at time of diagnosis; however, the risk score calculated in their study may only be most applicable within their institution.
Unfortunately, patients with UC and their families often do not have an accurate understanding of what their medical treatment may entail (16). Knowledge of the clinical factors identified in the present study will likely facilitate more accurate risk stratification at the time of diagnosis and throughout medical treatment. This will ultimately encourage earlier preoperative counseling and surgical referral. Children undergoing restorative proctocolectomy for UC have a high quality of life after surgery (17–19), and consideration of earlier surgical intervention for children with UC has been reported (20).
One limitation of the present study is that our analysis did not incorporate more global measurements of disease severity such as the Pediatric Ulcerative Colitis Activity Index (PUCAI). First validated in 2007 (21), the PUCAI has been demonstrated to aid in determining timely introduction of second-line therapy in severe acute UC (22). Furthermore, the Food and Drug Administration endorsed the PUCAI as a substitute to endoscopic evaluation for the primary outcome measure in a pediatric clinical trial evaluating a 5-ASA regimen. The PediIBDC dataset used for the present study predates the PUCAI, and included only dichotomized information on stooling patterns as opposed to the numerical gradations found in the PUCAI. Noninvasive, global clinical assessments such as the PUCAI are likely to have high predictive capabilities when determining need for colectomy in children with UC. Looking ahead toward future uses of the PediIBDC, now with 18 centers throughout the United States, Austria, and Hungary submitting data, a richer source of data may lead to further findings.
Unfortunately, like much of the literature examining clinical predictors of colectomy in children with UC, the present study is also limited by the relatively low number of children undergoing colectomy in the cohort. This limits the external validity of our predictors and those identified by previous groups, which may explain why there is notable variation between studies. Our results highlight the fact that even within centers of excellence in pediatric IBD care, there are relatively low numbers of colectomies performed overall. This makes identifying reliable surgical predictors a challenge for anyone examining health care utilization for children with UC and highlights the increased need for large, transnational, multiinstitutional collaborations.
In conclusion, children with a more severe disease burden are more likely to undergo surgical management of their UC. Specifically, children presenting with clinical and laboratory evidence of malnutrition, those with a first-degree family member with UC, and those treated with calcineurin inhibitors have an increased likelihood of undergoing surgery. Early identification and recognition of these factors should be used to facilitate preoperative counseling and patient-based planning for medical and surgical care of children with UC.
The authors thank the Pediatric Inflammatory Bowel Disease Foundation for sponsoring the Consortium and this work.
1. Gower-Rousseau C, Dauchet L, Vernier-Massouille G, et al. The natural history of pediatric ulcerative colitis: a population-based cohort study. Am J Gastroenterol
2. Baldassano RN, Piccoli DA. Inflammatory bowel disease in pediatric and adolescent patients. Gastroenterol Clin North Am
3. Turner D, Mack D, Leleiko N, et al. Severe pediatric ulcerative colitis: a prospective multicenter study of outcomes and predictors of response. Gastroenterology
4. Shiga H, Takagi S, Inoue R, et al. What determines the later clinical course of patients who do not undergo colectomy at the first attack? A Japanese cohort study on ulcerative colitis. Digestion
5. Watson S, Pensabene L, Mitchell P, Bousvaros A. Outcomes and adverse events in children and young adults undergoing tacrolimus therapy for steroid-refractory colitis. Inflamm Bowel Dis
6. Randall J, Singh B, Warren BF, et al. Delayed surgery for acute severe colitis is associated with increased risk of postoperative complications. Br J Surg
7. Gupta N, Cohen SA, Bostrom AG, et al. Risk factors for initial surgery in pediatric patients with Crohn's disease. Gastroenterology
8. Heyman MB, Kirschner BS, Gold BD, et al. Children with early-onset inflammatory bowel disease (IBD): analysis of a pediatric IBD consortium registry. J Pediatr
9. Russell RK, Satsangi J. IBD: a family affair. Best Pract Res Clin Gastroenterol
10. Laharie D, Debeugny S, Peeters M, et al. Inflammatory bowel disease in spouses and their offspring. Gastroenterology
11. Lees CW, Barrett JC, Parkes M, Satsangi J. New IBD genetics: common pathways with other diseases. Gut
12. Vind I, Riis L, Jess T, et al. Increasing incidences of inflammatory bowel disease and decreasing surgery rates in Copenhagen City and County, 2003–2005: a population-based study from the Danish Crohn colitis database. Am J Gastroenterol
13. Tremaine WJ, Timmons LJ, Loftus EV Jr, et al. Age at onset of inflammatory bowel disease and the risk of surgery for non-neoplastic bowel disease. Aliment Pharmacol Ther
14. Moore JC, Thompson K, Lafleur B, et al. Clinical variables as prognostic tools in pediatric-onset ulcerative colitis: a retrospective cohort study. Inflamm Bowel Dis
15. Hyams J, Markowitz J, Lerer T, et al. The natural history of corticosteroid therapy for ulcerative colitis in children. Clin Gastroenterol Hepatol
16. Baars JE, Siegel CA, Kuipers EJ, et al. Patient's perspectives important for early anti-tumor necrosis factor treatment in inflammatory bowel disease. Digestion
17. Carmon E, Keidar A, Ravid A, et al. The correlation between quality of life and functional outcome in ulcerative colitis patients after proctocolectomy ileal pouch anal anastomosis. Colorectal Dis
18. Hahnloser D, Pemberton JH, Wolff BG, et al. Results at up to 20 years after ileal pouch-anal anastomosis for chronic ulcerative colitis. Br J Surg
19. Krausz MM, Duek SD. Restorative proctocolectomy with ileal pouch-anal anastomosis for ulcerative colitis and familial adenomatous polyposis: twenty years follow-up in 174 patients. Isr Med Assoc J
20. Nagar H, Rabau M. The importance of early surgery in children with ulcerative colitis. Isr Med Assoc J
21. Turner D, Otley AR, Mack D, et al. Development, validation, and evaluation of a pediatric ulcerative colitis activity index: a prospective multicenter study. Gastroenterology
22. Turner D, Hyams J, Markowitz J, et al. Appraisal of the pediatric ulcerative colitis activity index (PUCAI). Inflamm Bowel Dis