Using faecal elastase-1 to screen for chronic pancreatitis in patients admitted with acute pancreatitis

Abstract

Background: Patients presenting with acute pancreatitis may have co-existing chronic pancreatitis, the accurate diagnosis of which would potentially guide appropriate management. Gold standard tests are often invasive, costly or time-consuming, but the faecal elastase-1 assay has been shown to be comparatively accurate for moderate and severe exocrine deficiency. This study aimed to evaluate fecal elastase-1 concentration [FE-1] against clinical criteria for chronicity in an acute setting. Patients and methods: [FE-1] was performed on patients admitted with acute onset of epigastric pain and a serum lipase at least three times the upper limit of normal. Clinical diagnosis of chronic pancreatitis was defined by the presence of specific clinical, pathological or radiological criteria. A [FE-1] value of <200 µg/g was similarly considered indicative of chronic exocrine insufficiency. Thus a 2×2 table comparing [FE-1] and clinical diagnosis was constructed. Results: After exclusion of liquid stool specimens, 105 stool specimens from 87 patients were suitable for [FE-1] determination. [FE-1] was evaluated against the clinical diagnosis of chronic pancreatitis, initially for the whole sample, and then after exclusion of cases of moderate and severe acute pancreatitis (Ranson score >2). The latter analysis, based on an exocrine insufficiency threshold of 200 µg/g, yielded a sensitivity of 79.5%, specificity of 98.0%, positive predictive value of 96.9% and negative predictive value of 86.0%. Conclusion: [FE-1] is an accurate screening tool for underlying chronic exocrine insufficiency when taken in the course of a hospital admission for mild acute pancreatitis.

Introduction

The Marseille-Rome classification of 1988 ¹ defines pancreatitis as either acute or chronic, eliminating the previous categories of acute relapsing and chronic relapsing pancreatitis. Acute pancreatitis is an acute inflammatory process of the pancreas with variable involvement of other regional tissues or remote organ systems ². Histological and physiological changes are potentially reversible. By contrast, chronic pancreatitis is a chronic inflammatory process characterized by irreversible or progressive fibrosis of the pancreas ³. Clinical manifestations include chronic epigastric pain (continuous or with intermittent exacerbations) and variable degrees of exocrine insufficiency (malabsorption with steatorrhoea, weight loss) and endocrine insufficiency (diabetes). While the exact prevalence of chronic pancreatitis in various populations is difficult to determine for logistic reasons, the annual incidence of acute pancreatitis has been reported as anywhere between 4.8 and 24.2 per 100 000 ⁴. It thus represents a significant proportion of the admissions to digestive surgery units.

Despite the apparent dichotomy of disease definition, it is evident that the two processes may occur simultaneously in some patients. In any clinical population, a proportion of incident cases of acute pancreatitis, defined by the usual diagnostic criteria, may harbour evolving chronic disease ⁵. The latter diagnosis is in practice based on clinical supposition, and then often only when symptoms of malabsorption or diabetes are overtly manifest. Early recognition of patients with underlying chronic pancreatitis is nevertheless useful so as to tailor a differential management pathway compared to those with purely acute disease.

Direct diagnostic tests of exocrine pancreatic function, such as the secretin-caerulein test ⁶, secretin-cholecystokinin test ⁷ and the Lundh meal ⁸, tend to be expensive, invasive or inconvenient. They are thus of limited use in all but academic practice. In the last several years, measurement of the faecal concentration of pancreatic elastase-1 ([FE-1]) has emerged as a relatively simple and non-invasive method of detecting exocrine insufficiency ⁹. Additional potential advantages include specificity for pancreas ¹⁰, stability during intestinal transit, at 4–8°C for 72 hours and at −20°C for up to 12 months ¹⁰,¹¹,¹²,¹³, concentration in faeces compared to pancreatic juice ¹¹ and non-cross-reactivity with therapeutic enzyme supplements ¹⁴.

Measured by the ELISA method, [FE-1] has compared favourably in terms of diagnostic efficacy to other indirect tests of pancreatic exocrine function including faecal chymotrypsin estimation ⁹,¹⁴,¹⁵, the para-aminobenzoic acid test ¹³ and the pancreolauryl test ¹⁴,¹⁶. However, Lankisch et al. concluded that [FE-1] was not distinctly superior to faecal chymotrypsin estimation, particularly in mild or moderate forms of chronic pancreatitis ¹⁷. Similar conclusions were drawn when comparing [FE-1] to the direct Lundh test ¹⁸.

This study has thus set out to determine the diagnostic efficacy of [FE-1] in a purely clinical setting where it is applied to opportunistic detection of underlying chronic pancreatitis in patients presenting with acute pancreatitis.

Patients and methods

All patients admitted with acute pancreatitis to a single regional hospital were consecutively recruited as part of an ongoing prospective cohort study. Diagnosis was based on the generally accepted criteria of a threefold or greater increase in serum lipase and typical epigastric pain ¹⁹. The Cairns Base Hospital, in the far north of Queensland, Australia, serves a population of approximately 250 000 from an area of >500 000 square kilometres. Indigenous Australians (Aborigines and Torres Strait Islanders) constitute about 12% of the population ²⁰. A previous retrospective study in this population has shown the aetiology of pancreatitis to be alcoholic in 62% of cases and biliary in 18% ²¹. For the purposes of the present study, aetiology was defined as biliary if gallstones were noted on standard medical imaging, and alcoholic if there was a reported recent intake of 80 g or more of alcohol per day in the absence of other definable risk factors such as gallstones, hypertriglyceridaemia, and hypercalcaemia ²². Severity of the acute episode was graded according to Ranson's predictive scoring system ²³. An episode was considered mild with a score of 0–2, moderate with a score of 3–4 and severe with a score of 5–11.

Where possible, the first available stool specimen was obtained from consenting patients during the course of their admission. If no stool specimen or only a subsequent specimen was obtained, these were excluded from the data analysis for this study. Frankly diarrhoeal or urine-diluted stool specimens were also excluded, as these are known to give falsely low [FE-1] measurements ⁹. [FE-1] was measured by the ELISA method (Schebo® Biotech, Giessen, Germany) according to the manufacturer's instructions. Diagnostic performance of this test was assessed with respect to a clinical ‘gold standard’, assuming exocrine insufficiency to be present if [FE-1] was <200 µg/g. This threshold is based on validated receiver operator characteristic calculations ⁹,²⁴. For the clinical gold standard diagnosis, chronic pancreatitis was suspected if one or more of the following were present: a history of typical chronic epigastric pain with no other explanation, more than four admissions for acute pancreatitis in the previous 5 years (where alcohol was considered to be the aetiological factor), a history of steatorrhoea, a history of otherwise unexplained weight loss, or suggestive features seen on medical imaging (e.g. calcification, atrophy, ductal abnormalities, pseudocyst formation). Diabetes mellitus was not included as a diagnostic criterion due to the already high prevalence of type 2 disease in certain sections of the target population.

Subsequent 2×2 tables were constructed to illustrate true positives (TP), false positives (FP), false negatives (FN) and true negatives (TN). Diagnostic performance was assessed in terms of: sensitivity = TP/(TP + FN), specificity = TN/(TN + FP), positive predictive value (PPV) = TP/(TP + FP), negative predictive value (NPV) = TN/(TN + FN) and diagnostic accuracy=(TP + TN)/(TP + FP + FN + TN).

Results

Between April 2001 and May 2004, 105 suitable stool specimens, corresponding to the same number of admissions, were obtained from 87 patients at a median of 4 days (range 0–20) after the onset of symptoms. Twelve liquid stool specimens had been excluded from the analysis. Eighty-nine (84.8%) of the evaluable admissions were considered mild on the basis of a Ranson score of 2 or less. Aetiological attribution was alcoholic in 64 admissions (61.0%), biliary in 24 (22.9%) and other/unknown in 27 (25.7%). [FE-1] values ranged from 0 to 920 µg/g with a mean of 283 µg/g. In 41 (39%), [FE-1] was measured as <200 µg/g. Forty-one specimens (39%) were obtained from patients in whom there was clinical evidence of chronic pancreatitis based on the above diagnostic criteria. The frequencies of each of these are shown in Table I. Twenty stool specimens (48.8%) were from patients with more than one positive criterion, while only 15 (36.6%) showed evidence of the later manifestations of chronic pancreatitis (chronic pain, weight loss, steatorrhoea).

Table I. Diagnostic criteria for chronic pancreatitis.

Criterion	Number of patients	Percentage of diagnosis positive group
At least one criterion	41	100%
Recurrent admissions	26	63.4%
Positive imaging	20	48.8%
Chronic pain	9	22.0%
Weight loss	6	14.6%
Steatorrhoea	5	12.2%

Results of the initial diagnostic performance analysis are shown in Table II. When cases of moderate or severe acute pancreatitis were excluded from the analysis, the diagnostic performance of [FE-1] improved considerably, largely due to elimination of all but one of the original false positives (see Table III).

Table II. [FE-1] compared to clinical diagnosis: all cases.

[FE-1]	Clinical diagnosis positive	Clinical diagnosis negative	Totals
[FE-1] <200 µg/g	32	9	41
[FE-1] >200 µg/g	9	55	64
Totals	41	64	105

Sensitivity = 78%, specificity = 76.6%, positive predictive value (PPV) = 78%, negative predictive value (NPV) = 85.9%, diagnostic accuracy = 82.9%.

Table III. [FE-1] compared to clinical diagnosis: mild acute pancreatitis (Ranson score <3).

[FE-1]	Clinical diagnosis positive	Clinical diagnosis negative	Totals
[FE-1] <200 µg/g	31	1	32
[FE-1] >200 µg/g	8	49	57
Totals	39	50	89

Sensitivity = 79.5%, specificity = 98%, positive predictive value (PPV) = 96.9%, negative predictive value (NPV) = 86%, diagnostic accuracy = 89.9%.

Discussion

This study aimed to evaluate the efficacy of [FE-1] as an objective diagnostic tool in a practical clinical setting, rather than in comparison to another test of pancreatic exocrine function. It further represents a unique attempt to diagnose chronic pancreatitis in incident cases of acute pancreatitis at the time of, or just after, the acute attack. The criteria chosen for the clinical gold standard diagnosis are all well accepted features of chronic pancreatitis, and are no different to what would normally be used in clinical practice to make a provisional diagnosis. This is particularly the case where the aetiological attribution is largely alcoholic, as is indeed evidenced in the current sample population and in previous epidemiological work in the region ²¹.

As shown in Table II, the application of [FE-1] to all incident cases of acute pancreatitis yielded unspectacular results for the usual indices of diagnostic performance. These markedly improved after exclusion of cases with moderate or severe acute disease as predicted by Ranson's score. Such cases were responsible principally for false positive results, as seen in Table III. Exocrine function may indeed be impaired by transient acinar cell dysfunction or frank parenchymal necrosis. Using [FE-1], Boreham and Ammori ²⁵ have shown that pancreatic exocrine insufficiency occurs commonly in patients recovering from severe acute pancreatitis, and its severity correlates with the extent of necrosis documented on contrast-enhanced CT scan. Some cases of moderate or severe acute pancreatitis may nevertheless yield a normal [FE-1] given the relative stability of this enzyme in the gastrointestinal tract ¹². A stool specimen obtained early in the course of an admission may represent that which was in the colon just prior to the onset of the acute attack. However, the ileus often present in severe cases may lead to eventual degradation of the enzyme, thus also ‘falsely’ lowering the [FE-1] measurement. Indeed in this study, median time of first stool specimen from onset of symptoms was 3 days for mild cases (Ranson score <3) compared with 5.5 days for moderate or severe cases (Ranson score >2).

A number of considerations would recommend the use of [FE-1] as a cost-effective screening tool for chronic exocrine insufficiency in patients presenting with acute pancreatitis. Firstly, the assay itself is relatively inexpensive and easy to perform. Stool specimens can also be stored in a freezer at −20°C until a sufficient number has been accrued for a run. Secondly, with a specificity of 98% and positive predictive value of 96.9%, a positive result can generally be relied upon. Furthermore, PPV is optimized when the target screening population has a relatively high prevalence of the disease in question ²⁶. This would seem to be the case in patients presenting with attacks of acute pancreatitis, particularly where the predominant aetiological factor is alcohol, as in this sample population. Thirdly, although [FE-1] is demonstrably less reliable when dealing with moderate or severe cases of pancreatitis, the vast majority of incident cases are typically mild ², as is also seen in the current series. Lastly, patients with chronic pancreatitis, particularly when alcohol-related, often pose problems for compliance and follow-up. A non-invasive test such as [FE-1] measured once during the course of a hospital admission is thus more feasible than most of the other diagnostic tests currently available.

It should be noted that 39% of the sample population showed evidence of chronic exocrine insufficiency based of a [FE-1] <200 µg/g. Extrapolation of this prevalence to the overall target population must take into account the fact that an estimated 30% of incident cases were initially excluded due to unavailability of a stool specimen or delayed specimen collection (i.e. not the first one after symptom onset). It is nevertheless difficult to conceive a specific selection bias under these circumstances in favour of any particular clinicopathological subgroup.

What to do with a positive [FE-1] result remains open to the outcomes of ongoing basic and clinical research. It is conceivable that chronic pancreatitis diagnosed prior to the development of the end-stage manifestations of intractable pain, malabsorption and diabetes may be amenable to innovative secondary prevention strategies. A simple objective diagnostic test is in any case useful for defining patients suitable for inclusion in clinical trials. Positive screening results may also have specific epidemiological implications for the local target populations. Twelve (31.6%) of the 38 patients in this study with [FE-1] below 200 µg/g were indigenous Australians. This group of patients is thus over-represented for chronic pancreatitis when compared with the target population as a whole. Such a result warrants culturally orientated management strategies as well as further studies to elucidate the precise causative factors.

In conclusion, [FE-1] is a feasible and accurate screening test for chronic exocrine insufficiency in patients presenting with predicted mild acute pancreatitis. While all acute patients are amenable to screening, it is recommended that those with moderate or severe acute disease, particularly if the initial [FE-1] result is subnormal, should be re-screened after 2–3 months to assess for residual or permanent exocrine insufficiency. Similarly those with normal [FE-1] but clinical features suggestive of underlying chronic disease should undergo follow-up testing to detect deterioration in exocrine function before more severe clinical features manifest.