Article Text

Download PDFPDF

Predicting a biliary aetiology in paediatric acute pancreatitis
  1. Michael J Coffey1,
  2. Scott Nightingale2,3,
  3. Chee Y Ooi1,4
  1. 1School of Women's and Children's Health, Medicine, University of New South Wales, Sydney, New South Wales, Australia
  2. 2Department of Gastroenterology, John Hunter Children's Hospital, Newcastle, New South Wales, Australia
  3. 3Discipline of Paediatrics and Child Health, School of Medicine and Public Health, University of Newcastle, Newcastle, New South Wales, Australia
  4. 4Department of Gastroenterology, Sydney Children's Hospital Randwick, Sydney, New South Wales, Australia
  1. Correspondence to Dr Chee Y Ooi, Sydney Children's Hospital, School of Women's and Children's Health, High Street, Randwick, Sydney, NSW 2031, Australia; keith.ooi{at}unsw.edu.au

Abstract

Objective Acute biliary pancreatitis (ABP) is a common cause of pancreatitis which may require timely intervention. We aimed to identify routine laboratory parameters for early prediction of biliary aetiology in paediatric acute pancreatitis (AP).

Design A retrospective review of children with AP (January 2000–July 2011) was performed at two tertiary paediatric hospitals in New South Wales, Australia. Predictors of ABP using laboratory parameters (measured within 48 h of presentation) were determined using a multivariate logistic regression model and evaluated.

Results Of the 131 pancreatitis episodes reviewed, 21 (16%) were biliary-related. Raised serum lipase, alkaline phosphatase, γ-glutamyl transpeptidase (GGT), alanine aminotransferase (ALT) and aspartate aminotransferase were associated with biliary aetiology (all p<0.0016) on univariate analysis. In multiple logistic regression, serum GGT ≥40 U/L, ALT ≥150 U/L, and lipase ≥15×the upper limit of normal (ULN) were predictive of ABP. To improve clinical applicability, the predictive performance when at least two of the three ABP predictors (coined the ‘biliary pancreatitis triad’) were satisfied was evaluated. The triad performed with a specificity, sensitivity, positive predictive value and negative predictive value of 95, 89, 76 and 98%, respectively.

Conclusions The biliary pancreatitis triad of serum GGT ≥40 U/L, ALT ≥150 U/L and lipase ≥15×ULN within 48 h of presentation may be used as simple clinical predictors of ABP in children. Children with values falling below 2 or 3 of these thresholds are very unlikely to have AP due to a biliary cause.

  • Gastroenterology
  • General Paediatrics
View Full Text

Statistics from Altmetric.com

What is already known on this topic

  • Acute biliary pancreatitis is a common cause of acute pancreatitis (AP) in children and may require timely and specific endoscopic or surgical intervention.

  • Previous studies comparing biliary and non-biliary AP suggest potential utility of serum biochemical markers in predicting biliary aetiology.

What this study adds

  • This study presents a simple clinical tool which may be used to predict the probability of a child having acute biliary pancreatitis (ABP) within 48 h of presentation.

  • The ‘biliary pancreatitis triad’ of serum γ-glutamyl transpeptidase ≥40 U/L, alanine aminotransferase ≥150 U/L and lipase ≥15×upper limit of normal were significantly associated and predictive of ABP.

Introduction

Acute pancreatitis (AP) remains a poorly understood condition in children despite an increasing incidence and associated morbidity and mortality.1–4 The causes of AP in children are very diverse and vary considerably in proportion between reported series.1–3 5–13 Acute biliary pancreatitis (ABP) is one of the most common causes of AP in children, comprising of 10–30% of all cases.14 Biliary pancreatitis in children can be due to several causes including obstruction of the common bile duct (CBD) by gallstone(s), sludge, choledochal cyst, biliary and/or pancreatic tree anomalies, duodenal obstruction, parasites and endoscopic retrograde cholangiopancreatography (ERCP) complications.3 ,15 ,16

The diagnosis of AP in children and adults requires two of the following three parameters: (1) typical abdominal pain, (2) an elevated lipase/amylase ≥3 times (×) the upper limit of normal (ULN) laboratory range, and/or (3) consistent findings on imaging.17 ,18 In adults, the imaging modality of choice is CT. By contrast, due to concerns of radiation exposure, abdominal ultrasonography is often the imaging modality of choice in children. However, ultrasonography has been reported to only identify an abnormality that is consistent with the diagnosis of AP in about one-third to one-half of cases.7 ,8 ,10 ,19 Nevertheless, in patients with ABP, imaging may enable appropriate timely endoscopic or surgical intervention.

To our knowledge, there have only been two paediatric studies which have compared features of biliary and non-biliary cases of paediatric AP.16 ,20 The findings from these studies suggest potential utility of serum biochemical markers in predicting biliary aetiology. In both studies, serum aspartate aminotransferase (AST) and lipase were significantly elevated in biliary compared with non-biliary AP. However, neither of these studies evaluated a comprehensive list of routine laboratory markers, or determined cut-offs for these markers and assessed their performance. Our study aimed to identify early (within 48 h of hospital presentation) laboratory predictors of biliary pancreatitis in a cohort of paediatric patients.

Methods

Study population

A retrospective review (January 2000–July 2011) was performed in all patients admitted to the Sydney Children's Hospital Randwick (SCH) and John Hunter Children's Hospital (JHCH). Both hospitals are tertiary referral hospitals for their respective regions in the state of New South Wales, Australia. Patients were identified using health record diagnostic codes for pancreatitis in medical records, and also by a laboratory database search for serum lipase or amylase ≥3×ULN.

Patients <18 years old at time of presentation were eligible for inclusion if they had a diagnosis of AP or acute recurrent pancreatitis (ARP). AP was defined as abdominal pain not due to other causes, plus either elevated serum lipase or amylase ≥3×ULN and/or imaging evidence of pancreatitis.17 ,18 Complete resolution of pain and at least 1 month pain-free interval between episodes was required to be considered ARP. Each documented episode of ARP was analysed as a separate AP episode. Patients presenting with pain and elevation of serum pancreatic enzymes due to pseudocyst(s) were excluded. Subjects in this study have been previously reported in a separate paper evaluating predictors of severe AP.21

Demographic, clinical, laboratory and radiographic data were collected from the medical records of patients with a confirmed diagnosis of AP. Laboratory data within 48 h of initial hospital presentation was analysed, and relevant maximum or minimum values were recorded. If a patient presented to a peripheral hospital initially, data was included and analysed from the time of initial presentation to that institution. Unavailable data for a given parameter was recorded as missing.

Determination of ABP

ABP was defined by association with gallstones, biliary sludge or structural anomalies of the biliary tract (eg, choledochal cyst). For the purposes of this study, AP associated with congenital anomalies of the pancreatic tract (eg, pancreas divisum) were also included as ABP. Abnormalities were determined by examining ultrasonography, CT, ERCP and/or magnetic resonance cholangiopancreatography reports in the patient file. A diagnosis of sludge-induced pancreatitis required radiographic evidence of CBD dilatation associated with sludge in the biliary tract.

Statistical analysis

To compare biliary and non-biliary episodes of AP, continuous variables were evaluated using unpaired Student t tests or Mann–Whitney U tests depending on the normality of data distribution. Due to the different reference ranges for serum lipase and amylase at each hospital laboratory, values for lipase and amylase were presented as the ratio above the ULN. Each categorical variable was examined using Fisher's exact test. A p value <0.05 was considered statistically significant; with Dunn-Sidak correction applied to account for multiple comparisons. Parameters that had a statistically significant association with ABP underwent receiver-operating characteristic (ROC) analysis to determine optimal cut-offs. Cut-offs for univariate relationships were then examined using binary logistic regression analysis to determine significance (p<0.05) and calculate OR with 95% CI. Any parameter with more than 25% missing data was excluded and all significant parameters were tested for collinearity (with a variance inflation factor >2 considered problematic) before inclusion in a multivariate binary logistic regression model. An OR with 95% CI, specificity, sensitivity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (PLR), negative likelihood ratio (NLR), positive post-test probability (PPTP) and negative post-test probability (NPTP) were calculated for any significant parameter(s) within the model. All statistical calculations were performed in SPSS V.21.0 (SPSS, Chicago, Illinois, USA).

Results

Study population

A total of 131 AP episodes from 125 patients were identified from the two institutions. Twenty-one of the 131 (16%) AP episodes were associated with identified biliary causes. Of the 21 ABP cases, 11 (52%) were associated with gallstones, 5 (24%) were associated with a choledochal cyst (Type I or IVa), 3 (14%) were associated with sludge, and 2 (10%) were associated with pancreas divisum. Of the 11 patients with gallstones, 7/11 (64%) cases were associated with biliary dilatation (in the absence of any pancreatic duct dilatation); 2/11 (18%) patients underwent ERCP and 9/11 (82%) patients underwent a subsequent cholecystectomy. None of the cases identified with sludge and biliary dilatation had associated pancreatic duct dilatation or biliary strictures. Non-biliary cases included: 35 idiopathic (27%), 33 traumatic (25%), 15 drug-related (11%), 9 infective (7%), 9 metabolic (7%), 3 due to systemic illness (2%), 3 genetic (2%), and 3 other causes (2%). Of the 131 AP episodes, 20 episodes (15%) had no imaging results recorded, and of these only 3 (2%) were classified as idiopathic cases.

The demographic data for biliary and non-biliary AP episodes are summarised in table 1. There were no significant differences in age or weight z-scores between biliary versus non-biliary AP patients. There was a significant female predominance in the biliary cohort (16 (76%) vs 46 (42%); p=0.004).

Table 1

Demographic characteristics

Derivation of predictor(s) of ABP

Univariate analysis

Comparisons between biliary and non-biliary episodes of AP for routine laboratory parameters are shown in table 2. To account for multiple comparisons, the Dunn-Sidak correction lowered the p value for statistical significance from 0.05 to 0.0016. Peak serum lipase, alkaline phosphatase (ALP), γ-glutamyl transpeptidase (GGT), AST and alanine aminotransferase (ALT) values were all significantly higher in the biliary group (table 2). Using ROC analysis, a cut-off was determined for each of these parameters, optimising both specificity and sensitivity (figure 1).

Table 2

Laboratory parameters within 48 h from time of presentation

Figure 1

Receiver operator characteristic analysis. Cut-off values were chosen to optimise specificity and sensitivity. AUROC, area under the receiver operator characteristic curve.

Based on these cut-offs, univariate logistic regression analysis was used to assess each parameter (table 3A). All parameters with their defined cut-offs were statistically significant.

Table 3

Logistic regression analyses of potential predictors of acute biliary pancreatitis

Multivariate analysis

Tests for collinearity revealed a relation (variance inflation factor >2) between ALT and AST. As ALT appeared to be a better predictor of ABP on univariate logistic regression analysis, AST was excluded from further analysis. The remaining parameters were included in the final multivariate logistic regression model (table 3B). Serum GGT ≥40 U/L, ALT ≥150 U/L and lipase ≥15×ULN were associated with a statistically significant increased risk of ABP. Although there was no significant collinearity between GGT and ALP, ALP was found to be not statistically significant on multivariate analysis.

Performance of identified predictors

The performances of the identified predictors of ABP are presented in table 4A. Of note is the high NPV of GGT, ALT and lipase being 96, 93 and 94%, respectively. In our cohort where the pre-test probability of ABP was 16%, using serum GGT, ALT and lipase increased the PPTP to 53, 59 and 32%, respectively; and decreased the NPTP to 4% 7 and 6%, respectively. Of the 131 episodes in our cohort, the number of patients who did not have a serum GGT, ALT and lipase test within 48 h of presentation were: 9 (1 biliary), 7 (1 biliary) and 7 (3 biliary) patients, respectively. These cases were treated as missing data. The predictive performances of all combinations of the significant variables were also analysed (table 4B).

Table 4

Predictors of acute biliary pancreatitis

To improve clinical applicability, the predictive performance when at least two of the three ABP predictors were met (GGT ≥40 U/L, ALT ≥150 U/L and lipase ≥15×ULN) was determined (table 4C). In such instances, the OR (95% CI, p value) for ABP was 150.4 (26.8 to 843.1, p<0.001). It performed with specificity, sensitivity, PPV, NPV, PLR and NLR of 95%, 89%, 76%, 98%, 17.6 and 0.12, respectively. In our cohort with a pre-test probability of 16% for ABP, having at least two of the three predictors increased the PPTP to 76% and decreased the NPTP to 2%. Of the 131 episodes in our cohort, 14 cases (3 biliary) did not have all parameters (GGT, ALT and lipase) recorded within 48 h of presentation and were treated as missing data.

Discussion

In this study, we report the potential usefulness of serum GGT, ALT and lipase as early predictors of biliary aetiology in paediatric AP. For the purposes of this discussion, we refer to them as the ‘biliary pancreatitis triad’. We observed that children with serum concentrations of GGT ≥40 U/L, ALT ≥150 U/L and lipase ≥15×ULN, performed within 48 h of presentation, were significantly more likely to have ABP. There was a significant association with ABP when at least two of the three ‘biliary pancreatitis triad’ parameters were met (OR=150.4, p<0.001; specificity 95%, sensitivity 89%). Furthermore, patients who did not fulfil at least two of the three parameters were very unlikely to have ABP (NPV 98%). As the true prevalence of AP in the paediatric population is unknown and may vary geographically,1 ,2 ,22 we also presented likelihood ratios (which are independent of disease prevalence) to allow application to different populations and settings.

Serum GGT, AST and lipase are usually ordered for the work-up and diagnosis of AP. Thus, the triad is a potentially simple tool for screening patients who are unlikely to have ABP without additional cost. The ability of these parameters to predict ABP within 48 h of presentation can provide clinicians with important aetiological information early in the course of disease. A high index of suspicion for ABP may then prompt early appropriate use of imaging modalities and, thus, enable timely planning for any intervention required.

ABP is thought to be related to the obstruction of outflow of pancreatic secretions through the CBD and ampulla of Vater.23 Greater elevations in lipase observed in biliary (vs non-biliary) AP may be explained by the obstructive stasis of pancreatic juice. This finding concurs with previous studies in children16 ,20 and adults.24–27 Similarly, elevations in liver transaminases (ALT and AST) have been reported elsewhere, with hepatic injury secondary to obstruction postulated. Choi et al16 demonstrated higher levels of both serum ALT and AST among patients with ABP, while Ma et al20 reported serum AST as a biomarker of ABP but did not include ALT in their analysis. We found ALT to be a better predictor than AST (table 3A), with the latter removed from multivariate regression analysis due to collinearity. In this study, raised serum GGT was also predictive of ABP. Although elevations in serum GGT was not unexpected in ABP due to obstruction of the biliary tract,28 this has not been reported in children with ABP. In fact, both previous studies in paediatric ABP did not report serum GGT. Consistent with our study, Choi et al16 reported higher serum ALP, another marker of biliary obstruction/disease, in biliary compared with the non-biliary pancreatitis.28 However, we found serum ALP to be not statistically significant after multivariate regression analysis. Serum GGT may be a better marker of biliary tract disease than ALP in children as compared with adults,26 ,27 particularly in normal growing children and rapidly growing adolescents, since elevated ALP levels may be of bone origin. In addition to their biologic plausibility, the triad of significant parameters were identified after robust statistical methods including Dunn-Sidak correction to account for multiple comparisons.

The proportion of ABP cases in our cohort was 16%, which is in accordance with previous reports of 10–30%.14 The frequency of other causes for AP were consistent with previous paediatric reports.14 Our definition criteria for ABP was also in accordance with previous paediatric studies.16 ,20 In our study, there were significantly more women with ABP compared with males (76% vs 42%, respectively; p=0.004). A higher percentage of females with ABP was also reported in the two previous paediatric studies,16 ,20 however these were not significant. Similarly, there is a female predominance in adults with ABP,26 ,27 with oestrogen implicated in gallstone formation in females.29 ,30

This study has several limitations. Since it is retrospective, not all patients had all blood tests performed within 48 h of presentation, and data collection was limited by the availability of medical records, resulting in incomplete data. Due to the relatively small sample size, other potentially significant parameters may not have reached statistical significance. We were unable to assess the influence of obesity31 since height data was infrequently recorded and, thus, body mass index could not be calculated. Additionally, patients with an underlying biliary aetiology may not have been identified as such, depending on whether imaging was performed and the type of imaging used. Since there is wide variability in aetiological distribution of ABP between various series, the generalisability of our specific cut-offs may be limited. Thus, further large multicentre prospective studies involving subjects from different demographics and regions are needed to properly validate our findings, as well as to determine other potential markers to predict ABP.

In conclusion, we have identified the potential utility of the ‘biliary pancreatitis triad’ of serum GGT, ALT and lipase as simple early predictors of ABP in children: those who do not fulfil at least two of serum GGT ≥40 U/L, ALT ≥150 U/L and lipase ≥15 × ULN within 48 h of presentation are unlikely to have ABP.

Acknowledgments

The authors would like to thank: Professor Andrea Rita Horvath and Keith Westbury, South Eastern Area Laboratory Services; Marcin Pasternak, Jacqueline Chao and Filip Tota, Prince of Wales Medical Records; and A/Prof Huy Tran, Hunter Area Pathology Service.

References

View Abstract

Footnotes

  • Contributors MJC: data acquisition, analysis and interpretation, statistical analysis and drafting of manuscript. SN: analysis and interpretation, statistical analysis and critical revision of manuscript. CYO: study conception and design, analysis and interpretation, statistical analysis, critical revision of manuscript and study supervision.

  • Competing interests None.

  • Ethics approval This study was approved by the institutional review boards (South Eastern Sydney Human Research Ethics Committee (10/188) and Hunter New England Human Research Ethics Committee (11/02/16/5.07)).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.