Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Aug 1.
Published in final edited form as: Clin Gastroenterol Hepatol. 2018 Jan 31;16(8):1220–1228.e4. doi: 10.1016/j.cgh.2018.01.027

Diagnostic Performance of Measurement of Fecal Elastase-1 in Detection of Exocrine Pancreatic Insufficiency – Systematic Review and Meta-analysis

Rohini R Vanga 1, Aylin Tansel 1, Saad Sidiq 2, Hashem B El-Serag 1,3,4, Mohamed Othman 1
PMCID: PMC6402774  NIHMSID: NIHMS937689  PMID: 29374614

Abstract

Background & Aims:

Tests to quantify fecal levels of chymotrypsin like elastase family member 3 (CELA3 or elastase-1) in feces are widely used to identify patients with exocrine pancreatic insufficiency (EPI). However, the diagnostic accuracy of this test, an ELISA, is not clear. We performed a systematic review and meta-analysis to determine the accuracy of measurement of fecal elastase-1 in detection of EPI.

Methods:

We searched PubMed, Embase, and reference lists for articles through November 2016 describing studies that compared fecal level of elastase-1 with results from a reference standard, direct method (secretin stimulation test), or indirect method (measurement of fecal fat) for detection of EPI. Sensitivity and specificity values were pooled statistically using bivariate diagnostic meta-analysis.

Results:

We included total of 428 cases of EPI and 673 individuals without EPI (controls), from 14 studies, in the meta-analysis. The assay for elastase-1, compared to secretin stimulation test, identified patients with pancreatic insufficiency with a pooled sensitivity value of 0.77 (95% CI, 0.58–0.89) and specificity value of 0.88 (95% CI, 0.78–0.93). In an analysis of 345 cases of EPI and 312 controls, from 6 studies, the fecal elastase-1 assay identified patients with EPI with a pooled sensitivity value of 0.96 (95% CI, 0.79–0.99) and specificity value of 0.88 (95% CI, 0.59–0.97), compared to quantitative fecal fat estimation. In patients with low pre-test probability of EPI (5%), the fecal elastase-1 assay would have a false-negative rate of 1.1% and a false-positive rate of 11%, indicating a high yield in ruling out EPI but not in detection of EPI. In contrast, in patients with high pre-test probability of EPI (40%), approximately 10% of patients with EPI would be missed (false negatives).

Conclusion:

In a systematic review and meta-analysis of studies that compared fecal level of elastase-1 for detection of EPI, we found that normal level of elastase-1 (above 200 mcg/g) can rule out EPI in patients with a low probability of this disorder (such as those with irritable bowel syndrome with diarrhea). However, in these patients, an abnormal level of elastase-1 (below 200 mcg/g) has a high false-positive rate.

Keywords: pancreas, enzyme, steatorrhea, exocrine pancreatic insufficiency, proteolysis

BACKGROUND

Progressive depletion of pancreatic acinar cells leads to exocrine pancreatic insufficiency (EPI), defined as an insufficient amount of pancreatic enzymes resulting in fat malabsorption [1]. Steatorrhea generally does not occur until lipase secretion is reduced by > 90% [2]. Because EPI is a frequent cause of diarrhea of obscure origin [3] and can be effectively treated with enzyme supplementation, the ability to accurately diagnose EPI is critical.

Although direct hormone-stimulated pancreas function tests remain the gold standard for the assessment of exocrine function of pancreas, these tests involve patient discomfort, cumbersome equipment, complication risks, non-standardized test protocols, and are not widely available [4]. Testing can be uncomfortable for patients due to placement of a Dreiling tube through the mouth, intravenous administration of secretin or cholecystokinin and periodic collection of pancreatic secretions from duodenum. A simplified version of this test called endoscopic secretin pancreatic function test has been developed but its diagnostic utility for EPI has not been proven [5, 6]. Clinicians frequently use empiric trial of pancreatic enzyme replacement and use patient reported reduction in steatorrhea to establish EPI diagnosis [7]. However, this is not ideal due to availability of different formulations of enzyme replacement, cost of medications, and patient compliance. For these reasons, identification of accurate, noninvasive biomarkers of disease activity is a priority.

Noninvasive (tubeless) pancreatic function tests include fecal chymotrypsin or elastase assays, 72-hr fecal fat, bentiromide (NBT-PABA), or fluorescein dilaurate testing [8, 9]. Fecal chymotrypsin assays are limited in that chymotrypsin is prone to proteolytic degradation and the assay cannot differentiate the exogenous chymotrypsin found in pancreatic enzyme supplements [10]. Quantitative (72-hr) fecal fat estimation is considered the gold standard among the indirect pancreatic function tests [11, 12]. However, this test measures global fat malabsorption (which includes EPI), is cumbersome, and is generally not well tolerated by patients.

Pancreatic elastase-1 is highly stable throughout the intestinal tract owing to a lack of relevant proteolytic degradation [13]. The fecal elastase-1 (FE-1) assay is an enzyme-linked immunosorbent assay (ELISA) that specifically measures pancreatic elastase-1, providing the opportunity to assess pancreatic function even in the presence of simultaneous enzyme supplementation. A recent population-based study in Norway reported prevalence of EPI as 11.5% among 914 participants; EPI was defined by FE-1 level <200 μg/g[14]. In another clinic-based study from England, the prevalence of EPI diagnosed also by low FE-1 (<200 μg/g) in 314 patients with chronic diarrhea who satisfied the Rome II criteria for diarrhea-dominant irritable bowel syndrome was 6.1%[15]. Furthermore, EPI prevalence of 4.4% (diagnosed by FE-1<200 μg/g) was documented in 90 patients who had serological and histological evidence of celiac disease in Slovenia; interestingly, the pancreas appeared normal on MRI imaging in all patients diagnosed with EPI in this study [16]. EPI was diagnosed (based on FE-1<200 μg/g) in 15% of asymptomatic recovered alcoholics in a prospective study from Brazil [17]. Therefore, EPI is likely to be common, and FE-1 may be an appropriate screening test.

The commercially available ELISA kit uses two specific monoclonal antibodies against human elastase 1 to apply on a spot fecal sample [18, 19]. Immunoassay for FE-1 is relatively inexpensive, requires < 1 gm of stool and is covered by most commercial insurance carriers [8]. The assay is not affected by diet or fasting status. Collected sample is stable for up to 14 days if kept refrigerated. Sample should not be watery or diluted. However, FE-1 tests are known to have assay variability, which could influence the accuracy of the test [20] [21]. Despite the widespread use of FE-1 in screening for EPI, the diagnostic accuracy of FE-1 is not clear.

We aimed to determine the diagnostic accuracy of FE-1 in detecting patients with EPI, and to evaluate effectiveness of FE-1 in differentiating mild, moderate and severe EPI. Our secondary aim was to evaluate diagnostic accuracy in clinically relevant subgroups.

MATERIALS AND METHODS

Search strategy

We searched PubMed and Embase from inception through November 2016 for relevant studies using a combination of MeSH terms and keywords according to PRISMA guidelines [22]. The search terms were (Pancreatic elastase[tiab] OR fecal elastase[tiab] OR faecal elastase[tiab] OR pancreatic function tests[mesh:noexp]) AND (exocrine pancreatic insufficiency [mesh: noexp] OR pancreatic insufficiency [tiab]). We did not use any search restrictions. Two authors (RV and SS) independently screened studies for possible inclusion in the review by reading the titles and abstracts. We retrieved the full text of the references that seemed to satisfy our protocol inclusion criteria. We limited studies to English, but considered studies in other languages if an English abstract was provided, sufficient data were provided and met inclusion criteria. Two authors reviewed abstracts and full text of the publication, and excluded non relevant studies. All disagreements in the screening and reviewing process were discussed and reviewed by a third author (AT). Eligible studies published as abstracts only were included only if they met inclusion criteria and sufficient data were available for analysis.

Types of studies and participants

We included studies (randomized clinical trials, cohort, case-control, or cross-sectional studies) that evaluated the diagnostic accuracy of FE-1 for the assessment of EPI. Study participants could have been pediatric or adult patients with EPI confirmed with secretin-cerulein test (SCT) or quantitative (72-hr or 24-hr) fecal fat estimation. Studies had to include a control group of patients (healthy or non-pancreatic disease controls) with normal pancreatic function. Studies needed to compare FE-1 results against either SCT or quantitative fecal fat analysis as the reference standards. A normal SCT was defined as fluid secretion > 67 mL/30 min, bicarbonate concentration >70 mol/L, bicarbonate output >6.5 mol/30 min after secretin administration, and amylase output >12,000 U/30 min, lipase output >21,000 U/30 min after cerulean administration[23]. Exocrine pancreatic function was evaluated as normal or abnormal according to the result of the SCT. Those with abnormal SCT, quantitative fat estimation was performed only in some studies (Table 1) to further stratify EPI into mild, moderate and severe as described above. This meant studies that performed SCT did not need to have quantitative fecal fat test or vice-versa. There were only 6 studies in the meta-analysis that compared FE-1 with quantitative fecal fat estimation. Therefore, we separated the studies based on comparison of FE-1 with SCT or FE-1 with quantitative fecal fat estimation.

Table 1:

FE-1 vs secretin stimulation test; summary of studies included in the meta-analysis.

Study (ref) Year, Country Design Etiology of EPI FE-1 cut off Adults Age ≥ 18 (Yes/no) Children Age < 18 (yes/no) Cases (n) Age (Mean ± SEM Or Median (range)) Gender (M/F) Proportion of cases Healthy Controls (n) Age (Mean ± SEM Or Median (range)) Gender (M/F) Disease Controls (n) Age (Mean ± SEM or Median (range)) Gender (M/F)
Mild EPI Moderate EPI Severe EPI
Stein [10] 1996, Germany Case-control CP, CF, Pancrea tectomy 175 Yes Yes 54 - - 7/29 - 22/29 53 - - 57 - -
Loser [9] 1996, Germany Case-control CP 200 Yes No 44 42.5 ± 3.5 30/14 8/44 14/44 22/44 50 27.4 ± 0.8 26/24 35 52 ± 2.5 22/13
Amann [26] 1996, United States Case-control CP 200 Yes No 14 48 (26–57) 6/8 7/14 - 7/14 15 38 (28–67) 4/11 7 49 (28–63) 0/7
Katschinski [27] 1997, Germany Case-control CP 200 Yes No 11 26–70 7/4 - - - 10 23–26 0/10 12 26–70 8/4
Soldan [33] 1997, Germany Case-control CF 200 Yes Yes 16 17.3 (8–25.3) 9/7 1/16 - 15/16 23 5.7 (0.8–13.9) 17/6 - - -
Loser [34] 1998, Germany Case-control CP 200 Yes Yes 26 48.3 ± 5 18/8 13/26 - 13/26 27 30.2 ± 1.4 16/11 25 55 ± 2.5 16/9
Lankisch [35] 1998, Germany Cohort CP 200 Yes No 30 - - 10/30 9/30 11/30 - - - 34 - -
Walkowiak [36] 1999, Finland Case-control CF 200 Yes Yes 28 11.2 ± 4.1 17/11 4/28 4/28 20/28 55 9.5 ± 3.6 26/29 - - -
Luth [28] 2001, Germany Cohort CP 200 Yes No 62 - - 23/62 14/62 25/62 - - - 65 - -
Takeda [29] 2002, Japan Case-control CP, AP, Pcyst, Pea 200 Yes No 65 - - - - - 131 - - - - -
Hahn [23] 2005, Germany Cohort CP 200 Yes No 11 - - - 5/11 6/11 - - - 20 - -
Naruse [37] 2006, Japan Case-control CP 200 Yes Yes 45 30–83 35/10 18/45 9/45 18/45 62 5–70 32/30 - - -
Hahn [30] 2008, Germany Cohort Type 1 DM 200 Yes No 11 47.9 ± 8.6 6/5 - 11/11 - - - - 22 47.5 ± 14.2 11/11
Wali [31] 2012, United States Cohort failure to thrive 200 No Yes 11 2.6 ± 1.8 5/6 - - - - 59 - 35/24
Open in a new tab

Mild EPI was defined as reduced output of one or more enzymes, bicarbonate concentration and fecal fat excretion normal, moderate EPI as reduced enzyme output and bicarbonate concentration, fecal fat excretion normal and severe EPI as reduced enzyme output and bicarbonate concentration plus steatorrhea. In the 72-hr fecal fat analysis, steatorrhea was defined as >7gm fat excretion per day using the Va de Camer method [11].

Data extraction and management

Two authors (RV and SS) independently extracted data for all included studies including (title, journal, year, publication status, and study design), number of participants included, baseline characteristics (adult or pediatric, average age, etiology of EPI), the reference standard test employed (secretin stimulation test or quantitative fecal fat), features of FE-1 and reference standard tests (including cutoff values and methods of collection), and the number of true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN). We excluded from analysis studies that did not provide sufficient data to determine TP, TN, FP, and FN.

Quality assessment

We used the Quality Assessment of Diagnostic Accuracy Studies tool (QUADAS 2) to assess the risk of bias in four domains regarding participant selection, index test, target condition, reference standard, and flow and timing [24]. Two authors (RV and SS) independently assessed the risk of bias of the included studies. In case of disagreement, we resolved by discussion with third author (AT).

Statistical analyses

We performed two separate analyses. The first model pooled studies using the secretin stimulation test as the reference test, and the second model pooled studies using quantitative stool fat as the reference test. For each, we determined sensitivity, specificity and positive and negative likelihood ratios (LR+ and LR-) with their 95% confidence intervals (CI). In general, LR+ > 10 indicates that the test has large effect on increasing the probability of the disease, LR+ between 5 to 10 indicates a moderate effect on increasing the probability of the disease and LR+ <5 indicates that the test has a small effect on increasing the probability of the disease. For LR-, test value < 0.1 indicates that the test has large effect on decreasing the probability of the disease, test value between 0.5 to 0.1 indicates a moderate effect on decreasing the probability of the disease and LR- >0.5 indicates that the test has a small effect on decreasing the probability of the disease. We also displayed sensitivity and specificity data in forest plots, and graphed study-specific estimates of sensitivity and specificity with 95% CI in the receiver operating characteristic (ROC) space. A bivariate diagnostic meta-analysis with random effects was used to pool estimates for sensitivity and specificity. We assessed the usefulness of FE-1 in clinical practice by using FP and FN rates. We assessed between-study heterogeneity using the I 2 statistic.

We performed several post-hoc sensitivity analysis which included i) the studies of adults only comparing FE-1 with secretin stimulation test, ii) the studies which have chronic pancreatitis only as the etiology of EPI, iii) only studies that used FE-1 200 μg/g as cut off, and iv) excluding abstracts.

Using the logit transformations for sensitivity and specificity, we also performed graphics to assess aspects of data and identify outliers with scatter plot and chi-squared plot. Finally, we performed Deeks funnel plot asymmetry test to investigate potential for publication bias by visual inspection of the patterns drawn from study data, where lack of symmetry should denote high risk of reporting bias. All statistical analyses were performed using STATA 13.1 (College Station, TX).

RESULTS

Results of the search

We identified a total of 616 studies (Figure 1). Of which, only twenty studies (19 papers and 1 abstract) that were published fulfilled the inclusion criteria and were included in the review (Tables 1 & 2). All 20 studies were observational. A total of eight different countries represented. Nine studies evaluated adults only, 10 studies evaluated both adults and children and 1 study evaluated children only. In FE-1 vs. secretin stimulation test analysis of 14 studies, data on 1101 patients were assessed (428 with EPI and 673 controls who were either patients with other gastrointestinal illness (n=247) or healthy volunteers (n=426)). Likewise, in FE-1 vs quantitative fecal fat estimation analysis of 6 studies, data on 657 patients were assessed (345 with EPI and 312 controls who were either patients with other gastrointestinal illness (n=122) or healthy volunteers (n=190)). The severity of EPI based on secretin stimulation test was reported in 9 studies for mild, 7 for moderate and 10 studies for severe EPI. The most common causes of EPI were chronic pancreatitis (61.9%) and cystic fibrosis (28.5%). In 18 studies, cut-off point for FE-1 was identified as 200 μg/g. In 2 studies, cut-off points of 175 μg/g and 218 μg/g were identified for FE-1 [10] [25].

Figure 1:

Figure 1:

Open in a new tab

PRISMA Flow Diagram.

Table 2:

FE-1 vs quantitative fecal fat; summary of studies included in the meta-analysis.

Study (ref) Year, Country Design Etiology of EPI FE-1 cut off Stool collection (72hror 24hr) Adults Age ≥ 18 Children Age< 18 Cases (n) Healthy Controls (n) Disease Controls (n)
Walkowiak [38] 2002 Poland Case-control CF 200 72hr Yes Yes 107 105 -
Symersky [25] 2004 Netherlands Case-control CP 218 24hr Yes yes 38 40 31
Halloran [39] 2011 UK Cohort Pancreatic resection 200 72hr Yes no 79 - 48
Benini [40] 2013 Italy Cohort CP, CF, AP 200 24hr Yes yes 50 - 32
Corroccio [41] 2001 Italy Cohort CF 200 24hr Yes yes 49 45 -
Hahn [30] 2008, Germany Cohort Type 1 DM 200 72hr Yes No 22 - 11
Open in a new tab

CP – Chronic Pancreatitis, CF – Cystic Fibrosis, AP – Acute Pancreatitis, CD – Celiac Disease, Pcyst – Pancreatic Pseudocyst, Pca – Pancreatic Cancer

Diagnostic accuracy of FE-1 vs Secretin Stimulation Test

In the 14 studies that reported FE-1 vs. secretin stimulation test, data on 1101 patients were assessed (428 with EPI and 673 controls who were either patients with other gastrointestinal illness (n=247) or healthy volunteers (n=426)). The pooled sensitivity and specificity estimates were 0.77 (95% CI 0.58–0.89) and 0.88 (95% CI 0.78–0.93), respectively (Figure 2). The pooled positive and negative likelihood ratios were 6.1 (95% CI 3.0–12.4) and 0.27 (95% CI 0.13–0.54) and the diagnostic OR was 23 (95% CI 6–84), respectively. The TP/TN/FP/FN of FE-1 reported by all studies included in our meta-analysis are presented in supplementary table 6.

Figure 2:

Figure 2:

Open in a new tab

Forest plot showing study-specific and mean sensitivity and specificity of FE-1 compared to secretin stimulation test for exocrine pancreatic insufficiency with corresponding heterogeneity statistics.

Diagnostic accuracy of FE-1 versus Secretin (direct) Stimulation Test for mild, moderate and severe EPI

There was considerable increase in sensitivity and negative likelihood ratio as EPI severity increased. From nine studies, patients with mild EPI were pooled and analyzed for diagnostic accuracy of FE-1 (Supplementary Table 1). The pooled sensitivity and specificity estimates were 0.49 (95% CI 0.29–0.70) and 0.92 (95% CI 0.85–0.96), respectively. The pooled positive and negative likelihood ratios were 6.4 (95% CI 3.0–13.46) and 0.54 (95% CI 0.35–0.83), respectively.

Seven studies that had patients with moderate EPI were analyzed for diagnostic accuracy of FE-1 (Supplementary Table 2). The pooled sensitivity and specificity estimates were 0.67 (95% CI 0.25–0.92) and 0.86 (95% CI 0.72–0.93), respectively. The pooled positive and negative likelihood ratios were 4.8 (95% CI 1.75–13.55) and 0.37 (95% CI 0.10–1.31), respectively.

Ten studies that reported severe EPI were analyzed for diagnostic accuracy of FE-1 (Supplementary Table 3). The pooled sensitivity and specificity estimates were 0.97 (95% CI 0.86–0.99) and 0.91 (95% CI 0.84–0.95), respectively. The pooled positive and negative likelihood ratios were 11.3 (95% CI 5.90–21.45) and 0.02 (95% CI 0.004–0.15), respectively.

Based on the estimates of prevalence of EPI from recent population based studies, we illustrate the clinical utility of FE-1 in the context of evaluating symptoms in patients with suspected EPI using false positive and false negative rates. If the prevalence of EPI in patients meeting criteria for IBS-D or chronic diarrhea is 5%, then based on our pooled sensitivity and specificity estimates, an abnormal FE-1 test (<200 μg/g) would result in FP rate of 11%, and TP rate of 3.8%. This may result in over prescription of PERT since FP/TP rate is 3:1; But when the test is normal (FE-1>200 μg/g), then FN rate is 1.1%, which is acceptable in ruling out EPI in this population. In contrast, in a high prevalence population, such as chronic diarrhea in patients with chronic pancreatitis or CF with suspected EPI prevalence of 40%, an abnormal FE-1 test (<200 μg/g) would result in FP rate of 7.2% and TP rate of 30.8%; however, for a normal test (FE-1>200 μg/g) FN rate is 9.2%. In this situation, a trial of PERT may be reasonable.

Diagnostic accuracy of FE-1 versus quantitative fecal fat estimate

In 6 studies that compared FE-1 vs. quantitative fecal fat estimate, data on 657 patients were assessed (345 with EPI and 312 controls who were either patients with other gastrointestinal illness (n=122) or healthy volunteers (n=190)). The pooled sensitivity and specificity estimates were 0.96 (95% CI 0.79–0.99) and 0.88 (95% CI 0.59–0.97), respectively (Supplementary Table 4). The TP/TN/FP/FN of FE-1 reported by all studies included in our meta-analysis are presented in supplementary table 7. The pooled positive and negative likelihood ratios were 8.4 (95% CI 1.8–38.1 and 0.04 (95% CI 0.006–0.29), respectively. These results were similar, with sensitivity of 0.98 (95% CI 0.82–0.99) and specificity of 0.84 (95% CI 0.48–0.97), when the study by Symersky et al. which used FE-1 cut off 218 μg/g was excluded from the pooled analysis. Diagnostic accuracy of FE-1 versus quantitative fecal fat estimate for mild, moderate and severe EPI could not be performed as some of the original included studies did not stratify patients based on the severity of EPI. Also, further analysis was limited as there were only 3 studies that used 72-hour fecal fat. Rest of the studies used 24-hour fecal fat.

Sensitivity analysis for FE-1 versus Secretin Stimulation Test

Various sensitivity analyses were performed as summarized in table 3. When the studies with subjects ≥ 18 years of age were included [9] [26] [27] [28] [29] [23], [30] the pooled sensitivity and specificity estimates were 0.75 (95% CI 0.63 – 0.85) and 0.85 (95% CI 0.70–0.93), respectively. After excluding the study by Takeda et al, which was published in abstract form only [29], the pooled sensitivity and specificity estimates were 0.75 (95% CI 0.54–0.88) and 0.88 (95% CI 0.79–0.94), respectively. The pooled sensitivity and specificity estimates of studies with FE-1 cut off value of 200 μg/g were 0.74 (95% CI 0.53–0.87) and 0.86 (95% CI 0.76–0.93), respectively. Lastly, the pooled analysis of 8 studies (243 cases vs. 330 controls) with chronic pancreatitis as the etiology of EPI, sensitivity and specificity were 0.73 (95% CI 0.56–0.85) and 0.89 (95% CI 0.79–0.95), respectively. Specifically, excluding the study by Hahn et al [30] that included type 1 diabetes and study by Wali et al [31] that included patients with failure to thrive did not change the results.

Table 3:

Diagnostic accuracy of FE-1 for exocrine pancreatic insufficiency

Sensitivity Specificity Positive LR Negative LR DOR
Sensitivity analysis la 0.75 (0.63, 0.85) 0.85 (0.70, 0.93) 5.19 (2.43, 11.08) 0.28 (0.18, 0.44) 18.23 (6.72, 49.46)
Sensitivity analysis 2a 0.75 (0.54, 0.88) 0.88 (0.79, 0.94) 6.64 (3.06, 14.42) 0.27 (0.13, 0.59) 23.81 (5.69, 99.62)
Sensitivity analysis 3a 0.74 (0.53, 0.87) 0.86 (0.76, 0.93) 5.6 (2.67, 11.74) 0.29 (0.14, 0.66) 18.95 (4.89, 73.32)
Sensitivity analysis 4a 0.73 (0.56, 0.85) 0.89 (0.79, 0.95) 7.15 (3.55, 14.41) 0.29 (0.16, 0.51) 24.28 (8.90, 66.21)
Sensitivity analysis 5b 0.98 (0.82, 0.99) 0.84 (0.48, 0.97) 6.46 (1.36, 30.49) 0.02 (0.001, 0.28) 288.25 (5.76, 14407.99)
Open in a new tab
a

FE-1 vs secretin stimulation test - sensitivity analysis 1: excluding studies with subjects < 18yrs old; Sensitivity analysis 2: excluding study that was published in abstract form; sensitivity analysis 3: excluding study with FE-1 cut off value of 175 μg/g; sensitivity analysis 4: excluding studies with etiology other than chronic pancreatitis ([10] [29] [30] [31] [33] [36])

b

FE-1 vs quantitative fecal fat estimate - sensitivity analysis 5: excluding study with FE-1 cut off value of 218 μg/g

Methodological quality of included studies

The quality of all included studies is described in supplementary table 5. Most were at high risk for bias. Statistical heterogeneity was also demonstrated, I 2 value was 90.16% (95% CI 86.19–94.14%).

Supplementary Figures 1–3 show qualitative assessment of studies with secretin stimulation tests as the control with scatter plot, chi-squared plot and Deeks’ funnel plot asymmetry test. There were no outliers and the Deeks’ funnel plot did not show significant asymmetry; all suggestive of absence of publication bias. Qualitative assessment of studies with quantitative fecal fat as the control also did not show outliers and did not show significant asymmetry. Deeks’ secretin stimulation test as control (p=0.71) and Deeks’ fecal fat as control (p=0.15).

DISCUSSION

This is the first study in English literature to perform a systematic review and meta-analysis of the diagnostic utility of FE-1 in EPI. Based on our review, the pooled sensitivity and specificity of FE-1 vs secretin stimulation test when all 16 studies were included was 0.77 (95% CI 0.58–0.89) and 0.88 (95% CI 0.78–0.93), respectively and diagnostic accuracy was 23 (95% CI 6–84). This diagnostic accuracy was maintained when several sensitivity analyses were performed as shown in Table 3.

Our study has some limitations. There was a small number of studies assessed in the final analysis of severity of EPI. Studies that employed secretin stimulation test as the gold standard may have had technical variations among studies, such as protocols for collecting pancreatic juices that were not described in sufficient detail in some of the studies. Furthermore, we identified several sources of heterogeneity, including small sample sizes, study population, disease spectrum and quality of reporting. Overall, most of the included studies used surrogate outcomes (e.g., stool fat) instead of patient reported outcomes. The meta analysis augments the indirect evidence presented by the different studies.

We performed several subgroup analysis controlling for these factors. Lastly, only two studies looked at sensitivity and specificity of FE-1 with cut off value of 100 μg/g, and therefore we could not perform meta-analysis because of the small number of studies. Our systematic review examined the utility of FE-1 for the diagnosis of EPI in pediatric and adult patients. Only twenty studies met our search criteria, which required comparison with a gold standard test, secretin stimulation test or quantitative fecal fat estimate. The findings were heterogeneous with I 2 value of 90.16% (95% CI 86.19–94.14%). Study sample sizes were small, affecting the precision of diagnostic accuracy. For example, Wali et al consisted of 11 cases and 59 controls, which resulted in sensitivity of 0.00 and specificity of 0.49 [31]. Furthermore, variation in FE-1 cut off was identified in two studies, Stein et al (175 μg/g) [10], and Symersky et al. (218 μg/g) [25].

In terms of clinical utility, the usefulness of the test relies on the underlying prevalence of EPI in the population as well as the intrinsic test characteristics (false positive and false negative rates, sensitivity and specificity). In a low prevalence population such as chronic diarrhea or IBS-D, FE>200 μg/g can rule out EPI with very low FN rate of 1.1%; however, FE<200 μg/g, may result in over prescription of PERT since FP/TP rate is 3:1. In a high prevalence population such as chronic diarrhea in patients with chronic pancreatitis or CF, FE<200 μg/g can rule in EPI with very low FP rate of 7.2%; however, FE>200 μg/g may result in higher FN rates. In this situation, a trial of PERT may be reasonable.

The pooled sensitivities of FE-1 for diagnosing mild, moderate and severe EPI were 0.47 (95% CI 0.29–0.70), 0.67 (95% CI 0.25–0.92) and 0.97 (95% CI 0.86–0.99), respectively. These estimates are similar to those reported in German by Seigmund et al. (abstract in English) reported the pooled sensitivity of 0.54, 0.75 and 0.95 for mild, moderate and severe EPI respectively and specificity of 0.79 [32]. Therefore, the ability of FE-1 to detect mild and moderate EPI is poor to modest which can result in false negative cases in few circumstances. False negative test has the consequence of missing diagnosis of EPI which could prevent early interventions and treatment. On the other hand, false positive results, may lead to unnecessary treatment with pancreatic enzymes with possible side effects and unnecessary increase in the cost of care. Additionally, assay variability might have implications in utilizing FE-1 test as a screening tool in patients with chronic diarrhea, because a single value may be misleading, leading to a false positive diagnosis of EPI.

Based on our results, FE-1 demonstrated similar high sensitivity and specificity to diagnose severe EPI when compared with both direct (0.97 (95% CI 0.86–0.99) and 0.91 (95% CI 0.84–0.95), respectively) and indirect tests (0.96 (95% CI 0.79–0.99) and 0.88 (95% CI 0.59–0.97), respectively). Most studies included in meta-analysis comprised of patients with CP (10 studies) and only few had CF (3 studies). Approximately 36% of patients with CP presented with diarrhea that lead to the diagnosis of EPI. This finding suggests that FE-1 could be used in identifying EPI in the setting of CP presenting with or without diarrhea and also possibly reducing the use of empiric therapeutic trial of pancreatic enzymes when these are not indicated.

This systematic review indicates that fecal elastase-1 is a potentially useful diagnostic tool. A normal FE-1 can rule out EPI in low pre-test probability scenario (for example, patients with suspected IBS with diarrhea), and abnormal FE-1 in high pre-test probability scenario (for example, patients with chronic pancreatitis) can rule in EPI; in contrast, an abnormal FE-1 in a low-pretest probability may falsely classify patients as having EPI, whereas a normal FE-1 in a high pre-test probability scenario may be a false negative. Larger studies are needed to further characterize the utility of this fecal biomarker and determine optimal applications in clinical practice.

Supplementary Material

Supp_Figure3

Supplementary Figure 3: Deeks’ funnel plot for studies with quantitative fecal fat as gold standard

Supp_Figures

Supplementary Figure 1: Scatter plot and Chi-squared plot for studies with secretin stimulation as gold standard.

Supplementary Figure 2: Deeks’ funnel plot for studies with secretin stimulation as gold standard.

Supp_Tables

Supplementary Table 1: Sensitivity and specificity of FE-1 vs secretin stimulation test in individual studies with mild exocrine pancreatic insufficiency

Supplementary Table 2: Sensitivity and specificity of FE-1 vs secretin stimulation test in individual studies with moderate exocrine pancreatic insufficiency

Supplementary Table 3: Sensitivity and specificity of FE-1 vs secretin stimulation test in individual studies with severe exocrine pancreatic insufficiency

Supplementary Table 4: Sensitivity and specificity of FE-1 vs quantitative fecal fat in individual studies with exocrine pancreatic insufficiency

Supplementary Table 5: Methodological quality summary and review authors’ judgements about each methodological quality item for each included study.

Supplementary Table 6: True positive, false positive, false negative and true negative rests of studies included in the meta-analysis of FE-1 vs secretin stimulation test

Supplementary Table 7: True positive, false positive, false negative and true negative results of studies included in the meta-analysis of FE-1 vs quantitative fecal fat

GRANT SUPPORT:

Aylin Tansel is supported by NIH T-32 5T32DK083266–07 grant. This work is funded in part by the Texas Digestive Disease Center NIH DK58338. Dr. El-Serag is also supported by NIDDK K24-04-107.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

CONFLICT OF INTEREST: No conflict of interest to declare.

References

  • 1.Keller J and Layer P, Human pancreatic exocrine response to nutrients in health and disease. Gut, 2005. 54 Suppl 6: p. vi1–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.DiMagno EP, Go VL, and Summerskill WH, Relations between pancreatic enzyme outputs and malabsorption in severe pancreatic insufficiency. N Engl J Med, 1973. 288(16): p. 813–5. [DOI] [PubMed] [Google Scholar]
  • 3.Schiller LR, et al. , Diagnostic value of fasting plasma peptide concentrations in patients with chronic diarrhea. Dig Dis Sci, 1994. 39(10): p. 2216–22. [DOI] [PubMed] [Google Scholar]
  • 4.Lankisch PG, Exocrine pancreatic function tests. Gut, 1982. 23(9): p. 777–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Erchinger F, et al. , Quantification of pancreatic function using a clinically feasible short endoscopic secretin test. Pancreas, 2013. 42(7): p. 1101–6. [DOI] [PubMed] [Google Scholar]
  • 6.Stevens T and Parsi MA, Update on endoscopic pancreatic function testing. World J Gastroenterol, 2011. 17(35): p. 3957–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Toouli J, et al. , Management of pancreatic exocrine insufficiency: Australasian Pancreatic Club recommendations. Med J Aust, 2010. 193(8): p. 461–7. [DOI] [PubMed] [Google Scholar]
  • 8.Dominguez-Munoz JE, et al. , Fecal elastase test: evaluation of a new noninvasive pancreatic function test. Am J Gastroenterol, 1995. 90(10): p. 1834–7. [PubMed] [Google Scholar]
  • 9.Loser C, Mollgaard A, and Folsch UR, Faecal elastase 1: a novel, highly sensitive, and specific tubeless pancreatic function test. Gut, 1996. 39(4): p. 580–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Stein J, et al. , Immunoreactive elastase I: clinical evaluation of a new noninvasive test of pancreatic function. Clin Chem, 1996. 42(2): p. 222–6. [PubMed] [Google Scholar]
  • 11.Van De Kamer JH, Ten Bokkel Huinink H, and Weyers HA, Rapid method for the determination of fat in feces. J Biol Chem, 1949. 177(1): p. 347–55. [PubMed] [Google Scholar]
  • 12.Fine KD and Ogunji F, A new method of quantitative fecal fat microscopy and its correlation with chemically measured fecal fat output. Am J Clin Pathol, 2000. 113(4): p. 528–34. [DOI] [PubMed] [Google Scholar]
  • 13.Sziegoleit A and Linder D, Studies on the sterol-binding capacity of human pancreatic elastase 1. Gastroenterology, 1991. 100(3): p. 768–74. [DOI] [PubMed] [Google Scholar]
  • 14.Rothenbacher D, et al. , Prevalence and determinants of exocrine pancreatic insufficiency among older adults: results of a population-based study. Scand J Gastroenterol, 2005. 40(6): p. 697–704. [DOI] [PubMed] [Google Scholar]
  • 15.Leeds JS, et al. , Some patients with irritable bowel syndrome may have exocrine pancreatic insufficiency. Clin Gastroenterol Hepatol, 2010. 8(5): p. 433–8. [DOI] [PubMed] [Google Scholar]
  • 16.Vujasinovic M, et al. , Exocrine pancreatic insufficiency, MRI of the pancreas and serum nutritional markers in patients with coeliac disease. Postgrad Med J, 2015. 91(1079): p. 497–500. [DOI] [PubMed] [Google Scholar]
  • 17.Mattar R, et al. , Comparison of fecal elastase 1 for exocrine pancreatic insufficiency evaluation between ex-alcoholics and chronic pancreatitis patients. Arq Gastroenterol, 2014. 51(4): p. 297–301. [DOI] [PubMed] [Google Scholar]
  • 18.Pezzilli R, et al. , The ELISA fecal elastase-1 polyclonal assay reacts with different antigens than those of the monoclonal assay. Pancreas, 2005. 31(2): p. 200–1. [DOI] [PubMed] [Google Scholar]
  • 19.Schneider A, et al. , Monoclonal versus polyclonal ELISA for assessment of fecal elastase concentration: pitfalls of a new assay. Clin Chem, 2005. 51(6): p. 1052–4. [DOI] [PubMed] [Google Scholar]
  • 20.Phillips IJ, et al. , Faecal elastase 1: a marker of exocrine pancreatic insufficiency in cystic fibrosis. Ann Clin Biochem, 1999. 36 (Pt 6): p. 739–42. [DOI] [PubMed] [Google Scholar]
  • 21.Meyts I, et al. , Variability of fecal pancreatic elastase measurements in cystic fibrosis patients. J Cyst Fibros, 2002. 1(4): p. 265–8. [DOI] [PubMed] [Google Scholar]
  • 22.Moher D, et al. , Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med, 2009. 6(7): p. e1000097. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Hahn JU, et al. , A new fecal elastase 1 test using polyclonal antibodies for the detection of exocrine pancreatic insufficiency. Pancreas, 2005. 30(2): p. 189–91. [DOI] [PubMed] [Google Scholar]
  • 24.Whiting PF, et al. , QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med, 2011. 155(8): p. 529–36. [DOI] [PubMed] [Google Scholar]
  • 25.Symersky T, et al. , Faecal elastase-I: helpful in analysing steatorrhoea? Neth J Med, 2004. 62(8): p. 286–9. [PubMed] [Google Scholar]
  • 26.Amann ST, et al. , Fecal pancreatic elastase 1 is inaccurate in the diagnosis of chronic pancreatitis. Pancreas, 1996. 13(3): p. 226–30. [DOI] [PubMed] [Google Scholar]
  • 27.Katschinski M, et al. , Duodenal secretion and fecal excretion of pancreatic elastase-1 in healthy humans and patients with chronic pancreatitis. Pancreas, 1997. 15(2): p. 191–200. [DOI] [PubMed] [Google Scholar]
  • 28.Luth S, et al. , Fecal elastase-1 determination: ‘gold standard’ of indirect pancreatic function tests? Scand J Gastroenterol, 2001. 36(10): p. 1092–9. [DOI] [PubMed] [Google Scholar]
  • 29.Takeda M, et al. , [Fecal elastase-1 test: clinical evaluation of a new noninvasive pancreatic function test]. Rinsho Byori, 2002. 50(9): p. 893–8. [PubMed] [Google Scholar]
  • 30.Hahn JU, et al. , Low fecal elastase 1 levels do not indicate exocrine pancreatic insufficiency in type-1 diabetes mellitus. Pancreas, 2008. 36(3): p. 274–8. [DOI] [PubMed] [Google Scholar]
  • 31.Wali PD, et al. , Comparison of fecal elastase-1 and pancreatic function testing in children. J Pediatr Gastroenterol Nutr, 2012. 54(2): p. 277–80. [DOI] [PubMed] [Google Scholar]
  • 32.Siegmund E, Lohr JM, and Schuff-Werner P, [The diagnostic validity of non-invasive pancreatic function tests--a meta-analysis]. Z Gastroenterol, 2004. 42(10): p. 1117–28. [DOI] [PubMed] [Google Scholar]
  • 33.Soldan W, Henker J, and Sprossig C, Sensitivity and specificity of quantitative determination of pancreatic elastase 1 in feces of children. J Pediatr Gastroenterol Nutr, 1997. 24(1): p. 53–5. [DOI] [PubMed] [Google Scholar]
  • 34.Loser C, et al. , Comparative clinical evaluation of the 13C-mixed triglyceride breath test as an indirect pancreatic function test. Scand J Gastroenterol, 1998. 33(3): p. 327–34. [DOI] [PubMed] [Google Scholar]
  • 35.Lankisch PG, et al. , Faecal elastase 1: not helpful in diagnosing chronic pancreatitis associated with mild to moderate exocrine pancreatic insufficiency. Gut, 1998. 42(4): p. 551–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Walkowiak J, Cichy WK, and Herzig KH, Comparison of fecal elastase-1 determination with the secretin-cholecystokinin test in patients with cystic fibrosis. Scand J Gastroenterol, 1999. 34(2): p. 202–7. [DOI] [PubMed] [Google Scholar]
  • 37.Naruse S, et al. , Fecal pancreatic elastase: a reproducible marker for severe exocrine pancreatic insufficiency. J Gastroenterol, 2006. 41(9): p. 901–8. [DOI] [PubMed] [Google Scholar]
  • 38.Walkowiak J, et al. , Fecal elastase-1 is superior to fecal chymotrypsin in the assessment of pancreatic involvement in cystic fibrosis. Pediatrics, 2002. 110(1 Pt 1): p. e7. [DOI] [PubMed] [Google Scholar]
  • 39.Halloran CM, et al. , Partial pancreatic resection for pancreatic malignancy is associated with sustained pancreatic exocrine failure and reduced quality of life: a prospective study. Pancreatology, 2011. 11(6): p. 535–45. [DOI] [PubMed] [Google Scholar]
  • 40.Benini L, et al. , Fecal elastase-1 is useful in the detection of steatorrhea in patients with pancreatic diseases but not after pancreatic resection. Pancreatology, 2013. 13(1): p. 38–42. [DOI] [PubMed] [Google Scholar]
  • 41.Carroccio A, et al. , Diagnostic accuracy of fecal elastase 1 assay in patients with pancreatic maldigestion or intestinal malabsorption: a collaborative study of the Italian Society of Pediatric Gastroenterology and Hepatology. Dig Dis Sci, 2001. 46(6): p. 1335–42. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supp_Figure3

Supplementary Figure 3: Deeks’ funnel plot for studies with quantitative fecal fat as gold standard

NIHMS937689-supplement-Supp_Figure3.tif (476.1KB, tif)
Supp_Figures

Supplementary Figure 1: Scatter plot and Chi-squared plot for studies with secretin stimulation as gold standard.

Supplementary Figure 2: Deeks’ funnel plot for studies with secretin stimulation as gold standard.

NIHMS937689-supplement-Supp_Figures.docx (24.9KB, docx)
Supp_Tables

Supplementary Table 1: Sensitivity and specificity of FE-1 vs secretin stimulation test in individual studies with mild exocrine pancreatic insufficiency

Supplementary Table 2: Sensitivity and specificity of FE-1 vs secretin stimulation test in individual studies with moderate exocrine pancreatic insufficiency

Supplementary Table 3: Sensitivity and specificity of FE-1 vs secretin stimulation test in individual studies with severe exocrine pancreatic insufficiency

Supplementary Table 4: Sensitivity and specificity of FE-1 vs quantitative fecal fat in individual studies with exocrine pancreatic insufficiency

Supplementary Table 5: Methodological quality summary and review authors’ judgements about each methodological quality item for each included study.

Supplementary Table 6: True positive, false positive, false negative and true negative rests of studies included in the meta-analysis of FE-1 vs secretin stimulation test

Supplementary Table 7: True positive, false positive, false negative and true negative results of studies included in the meta-analysis of FE-1 vs quantitative fecal fat

NIHMS937689-supplement-Supp_Tables.pdf (110.5KB, pdf)

RESOURCES