Are we missing pancreatic exocrine insufficiency in ‘at-risk’ groups? Prospective assessment of the current practice and yield of faecal elastase testing in patients with diabetes mellitus, HIV and/or high alcohol intake
Abstract
There is cumulative evidence that pancreatic exocrine insufficiency (PEI) is under-recognised and can occur in patients with ‘at-risk’ conditions. Thus, we aimed to assess the current practice and yield of requesting faecal elastase (FEL-1), an indicator of PEI, in patients with ‘at-risk’ conditions. We prospectively recruited patients attending secondary care clinics with diabetes mellitus (DM), people living with HIV (PLHIV) and inpatients admitted to hospital with high alcohol intake (HAI). All patients underwent testing with FEL-1. Those patients with PEI (FEL-1 <200 μg/g) were contacted and offered a follow-up review in gastroenterology clinic. In total, 188 patients were recruited (HAI, n=78; DM, n=64; and PLHIV, n=46). Previous FEL-1 testing had not been performed in any of the patients. The return rate of samples was 67.9% for patients with HAI, 76.6% for those with DM and 56.5% for those with PLHIV. The presence of PEI was shown in 20.4% of patients with DM, 15.4% of patients with PLHIV and 22.6% in those with HAI. Diarrhoea and bloating were the most reported symptoms in followed-up patients with low FEL-1 (31.8% and 22.7% of patients, respectively). Follow-up computed tomography (CT) scans in those patients with PEI identified chronic pancreatitis changes in 13.6% and pancreatic atrophy in 31.8% of patients. These results suggest that there is a lack of testing for PEI in ‘at-risk’ groups. Our findings also suggest that using FEL-1 to test for PEI in patients with DM, PLHIV and HAI has a significant impact, although further studies are required to validate these findings.
Introduction
Pancreatic exocrine insufficiency (PEI) is defined as a reduced production or loss of activity of pancreatic enzymes in the small bowel, resulting in maldigestion symptoms and long-term consequences, including weight loss, micronutrient deficiency, osteoporosis, sarcopenia and malnutrition.1 Pancreatic enzyme replacement therapy (PERT) can improve symptoms and quality of life in patients with PEI.2 Primary causes of PEI are diseases that cause irreversible pancreatic parenchymal or ductal changes, including chronic pancreatitis (CP), cystic fibrosis, pancreatic cancer and acute necrotising pancreatitis.3 However, there is evidence that PEI can be caused by, or associated with, ‘at-risk’ conditions. ‘At-risk’ conditions refer to causes other than primary conditions, including high alcohol intake (HAI), diabetes mellitus (DM), and people living with HIV (PLHIV).3
Early diagnosis of PEI can be challenging because symptoms are non-specific, and the early morphological changes associated with CP can be subtle. Steatorrhoea can be a sign of late-stage disease and develops after significant reduction in lipase secretion (>90%)4; therefore, relying on steatorrhoea as a main symptom can result in a delayed diagnosis and missing early stages of PEI.
Current guidelines recommend the use of faecal elastase-1 (FEL-1) testing for PEI.2 FEl-1 >200 μg/g in stools indicates a normal exocrine pancreatic function, concentrations between 100 and 200 μg/g indicate mild–moderate PEI, and a level <100 μg/g indicates severe PEI.5 A recent meta-analysis using FEL-1 for detecting PEI showed a sensitivity of 77% and a specificity of 88% compared with secretin stimulation tests. Compared with quantitative faecal fat, FEL-1 showed a sensitivity of 96% and a specificity value of 88%.6 FEL-1 has a sensitivity of 54–90% and a specificity of 80% in mild–moderate PEI.7–10 In severe PEI, FEL-1 sensitivity increases to 100% and specificity to 93% compared with a faecal fat collection as a gold standard.11
False positive FEL-1 results can occur when stool is mixed with urine or watery diarrhoea; therefore, patients should be advised to avoid contamination when providing a sample.12 Using FEL-1 to diagnose PEI is less accurate compared with ‘gold standard’ tests, such as faecal fat collection. However, FEL- 1 is widely available and considered an acceptable alternative. Furthermore, direct ‘gold standard’ tests can be challenging to patients and staff, time-consuming and expensive13–16.
The ‘at-risk’ conditions shown to be associated with PEI include patients with DM in an outpatient setting, PLHIV and patients admitted with HAI-related symptoms. HAI is an established risk factor for benign pancreatic diseases. Pancreatitis can affect up to 55–72% of patients with a history of HAI and alcohol-related liver disease.17 Morphological changes of the pancreas have been reported in patients with DM on radiological and histological assessment18; in addition, chronic inflammatory changes and pancreatic atrophy have been shown in autopsy studies in 11–19% of patients with DM.19 The estimated prevalence of PEI in patients with type 1 DM is 26–74%, with a lower prevalence in those with type 2 DM (12–36%).19,20 Chronic diarrhoea in PLHIV is common and is associated with a decreased quality of life, PEI has also been shown to be present in 32–45% of PLHIV.21,22
The ‘at-risk’ groups are usually cared for by different medical specialties; therefore, the level of awareness of PEI can be inconsistent. With the evidence available, current clinical practice needs to be reviewed to assess which patients are being tested; therefore, we performed an observational research study to assess the current practice and yield of requesting FEL-1 in patients with ‘at-risk’ conditions.
Methodology
Population
We undertook an observational research study by prospectively recruiting patients with DM and PLHIV attending outpatient clinics, and inpatients admitted with HAI resulting in withdrawal symptoms.
The HAI cohort was defined as those patients admitted to hospital with alcohol withdrawal symptoms and a history of alcohol excess (>14 units per week) for more than 6 months. Inpatients with HAI were recruited from the acute medical and gastroenterology wards. Patients with DM (both type 1 and 2) and PLHIV were prospectively recruited from DM clinics and infectious diseases clinics, respectively.
Instructions for obtaining a stool sample and kits were provided with a labelled pot, request form, kidney-shaped disposable tray, plastic pathological bag, a pair of gloves and a non-transparent cardboard bag. A follow-up call was planned after 4 weeks to report the results, or send a reminder to return the sample. Patients with PEI were contacted and offered a follow-up review in our gastroenterology clinic.
Faecal elastase test
All FEL-1 samples were analysed in our centre with a commercially available enzyme-linked immunosorbent assay (ELISA), which uses two monoclonal antibodies against specific epitopes of human pancreatic elastase (ScheBo-Tech, Wettenberg, Germany). FEL-1 <200 μg/g was considered diagnostic of PEI. Patients with previous FEL-1 testing were not excluded. Those who were unable to provide a sample during their hospital stay or who were discharged soon after recruitment had the option to return a sample to either the hospital or their GP.
Control group
To demonstrate that FEL-1 yield was not the result of poor specificity, the prevalence of low FEL-1 in our cohorts was compared with a historical control cohort of 95 individuals recruited from secondary care gastroenterology clinics in our centre without symptoms of diarrhoea, irritable bowel syndrome or weight loss.23,24 Given that this group of patients were recruited from the same centre, and the same test and assay were used, we considered this to be a satisfactory comparison, acknowledging that slight changes in Rome Criteria for irritable bowel and a possible change in referral pattern to secondary care might have occurred after this control group was recruited. The prevalence of low FEL-1 in this control group was reported as 0/95 (0%; 95% confidence interval (CI), 0.0–3.8) for <100 μg/g and 3/95 (3.2%, 95% CI 0.7–8.9) for FEL-1 <200 μg/g.23
Follow up
Patients with low FEL-1 were offered a follow-up clinic review, including bloods tests and a computed tomography (CT) scan.25 A final radiological diagnosis of CP was made if patients had a Cambridge classification of ≥3 on CT. A CT scan was not performed in patients with normal FEL-1 because of the risk of radiation and lack of indication for imaging. In addition, no CT scan of the pancreas had been performed for other clinical reasons within the previous year in the patient cohort with normal FEL-1.
Data collection
In addition to the FEL-1 result, other data collected included height, weight, body mass index (BMI), pack-year smoking history, amount of alcohol consumed per week, and onset of disease in DM and PLHIV. The CT scan of pancreas findings of follow-up patients with low FEL-1 were also collected.
Statistical analysis
Data are reported as median and range for continuous variables using the Mann–Whitney U test. Categorical data were compared using Fisher's exact test. The proportion of patients with PEI in each group was calculated using the ‘Exact’ CI. Statistical significance was indicated by p<0.05. Statistical analyses were performed using SPSS (IBM SPSS statistics version 25, Chicago, IL, USA). Given that our aim was to assess the yield of FEL testing that exists and the yield and return of FEL-1 samples, a formal power calculation was not performed.
Ethical approval
This study was reviewed and approved by our local NHS Research Ethics Committee (STH20142/IRAS239539).
Results
None of the 188 recruited patients had previously been tested for PEI (HAI, 78; DM, 64; PLHIV, 46). The return rate of samples were as follows: HAI, 67.9% (53/78); DM, 76.6% (49/64); and PLHIV, 56.5% (26/46).
The presence of PEI (FEL-1 <200 μg/g) in the different groups was as follows: DM, 10/49 (20.4%, 95% CI 10.2–34.3); PLHIV, 4/26 (15.4%, 95% CI 4.4–34.9); and HAI, 12/53 (22.6%, 95% CI 12.3–36.2). Severe PEI (FEL-1 <100 μg/g) was demonstrated in 5/59 (10.2%) of patients with DM, 2/26 (7.7%) of PLHIV, and 7/53 (13.2%) of those with HAI. The prevalence of PEI was significantly higher in all (at-risk) groups (HAI, p=0.0003; DM, p=0.001, and PLHIV, p=0.04) compared with the historical control group (Fig 1).
The median alcohol intake in the HAI group was 140 units per week in both patients with low or normal FEL-1. Heterogeneous changes of the liver suggesting cirrhosis on transabdominal ultrasound were found in 72.7% (8/11) of patients with PEI compared with 39% (16/41) with a normal FEL-1 level. Most patients were either a current or past smoker with a median pack-year smoking history of 9.5. Two out of eight followed-up patients had morphological changes suggestive of CP.
The median BMI of the DM cohort with low FEL-1 was 28.5 kg/m2 and 7/10 (70%) of patients were male. Median pack-year smoking was 20 in the PEI group compared with 12 in those with normal FEL-1. Cross-section images of DM with low PEI identified a prevalence of pancreatic morphological changes in 6/10 (60%).
All four PLHIV with low FEL-1 were male, with a median BMI of 27.4 kg/m2; follow-up CT scans identified pancreatic atrophy in two of these patients.
Analysis of the association of smoking with PEI was performed in two ways; the median of pack-year smoking was compared between those with PEI and without PEI using the Mann–Whitney U test (Table 1). The correlation between pack-year smoking with FEL-1 level was also analysed. No significant regression association was found between smoking and FEl-1 levels in: HAI (F=0.7, R2=0.015, p=0.7); DM (F=2.37, R2=0.05, p=0.13); PLHIV (F=0.19, R2=0.005, p=0.119); or all cohorts (F=0.75, R2=0.006, p=0.75).
Diarrhoea and bloating were the most common reported symptoms in all patients with a low FEL-1 who attended a follow-up appointment (7/22 (31.8%) and 5/22 (22.7%), respectively).
Discussion
Despite previous FEL-1 testing not being an exclusion criterion, none of the recruited 188 patients in our study had been tested for PEI previously. The study showed that FEL-1 testing had a significant yield in ‘at-risk’ groups, which was significantly higher compared with historical controls. It is common to encounter patients with DM or HAI, or PLHIV in clinical practice across different medical specialties; therefore, a case-finding practice is recommended to initiate early treatment and prevent consequences of maldigestion and malabsorption.
Alcohol is metabolised by the pancreas through oxidative and non-oxidative pathways; the resulting metabolites inhibit pancreatic stellate cell apoptosis, thus facilitating prolonged survival and causing fibrosis. Furthermore, alcohol increases precipitation of pancreatic juice and the formation of protein plugs inside the pancreatic duct.26 The risk of acute alcoholic pancreatitis is dose dependent, with an odds ratio (OR) of 1.4 for 25 g alcohol/day compared with OR of 3.6 for 100 g alcohol/day.27 The median weekly intake in our HAI group was 140 U (17.5 g), which is 10 times higher than the recommended alcohol consumption (currently is 14 U per week for men and women in the UK).28 The risk of alcoholic CP increased in a step-wise manner, from an OR of 1.58 for 25 g alcohol/day, OR of 2.51 for 50 g/day, OR of 3.97 for 75 g/day, to OR of 6.3 for 100 g/day.29 Malnutrition is common in patients with HAI, including those without liver disease.30,31 Fat loss and sarcopenia can develop with inadequate protein intake. In addition to reducing mineral and vitamin intake, alcohol inhibits the absorption of nutrients, including thiamine.
Several mechanisms have been postulated to explain the exocrine dysfunction in patient with DM, including autonomic neuropathy, the inhibitory role of glucagon on the exocrine function, and pancreatic atrophy.32–35 In a recent cross-sectional study comparing patients with longstanding type 1 DM to healthy controls, FEL-1 was significantly lower in those with DM and correlated with C-peptide. Another large multicentre study reported a PEI prevalence of 42% in DM, correlated with duration of the disease.36
In a study of 100 patients with PLHIV with a median disease duration of 173 months, 32 were found to have FEL-1 <200 μg/g with improvement in symptoms achieved in 75% of patients receiving PERT.22 However, the mechanism of PEI in PLHIV is not well understood. A post-mortem study of relatively young PLHIV found that up to 90% of cases had pancreatic morphological changes.37 This study suggested parenchymal destruction as a cause of PEI. Another study showed pancreatic morphological changes in 45% of PLHIV who underwent pancreatic assessment with magnetic resonance imaging (MRI).38 The current study showed 50% pancreatic morphological changes, although a definitive conclusion could not be reached because of the small number of patients involved, although it concurs with the findings of this previous study.
Most patients with low FEL-1 were overweight and 77% had BMI ≥25 kg/m2 for all cohorts; therefore, relying on a single BMI measurement for nutritional assessment is inadequate and could miss cases of malnutrition and opportunity for early intervention, especially in those with obesity. High BMI with increased visceral adiposity can cause fatty replacement of the pancreas and atrophy with consequences of exocrine insufficiency.39 Ten out of the 22 patients with low FEL-1 who attended follow-up clinics showed morphological changes in the pancreas, with pancreatic atrophy being the most common 70% (7/10). The risk association between PEI and an atrophic pancreas can be overlooked, because the finding of an atrophic pancreas on radiological reports might not carry the same clinical status as calcified chronic pancreatitis.
PEI symptoms are non-specific and most patients in our cohort with low FEL-1 had no report of steatorrhoea or weight loss, making it challenging to diagnose PEI by only relying on malabsorption symptoms. Therefore, a composite assessment for such patients is required based on symptoms, the presence of an ‘at-risk’ condition, FEL-1 result, and nutritional assessment.
Patients with PEI should be treated with PERT to avoid long-term consequences and improve their nutritional status. PERT has been shown to improve gastrointestinal symptoms, nutritional status and quality of life, including in patients with non-CP changes.40
A significant proportion of inpatients with HAI did not return a sample (32.1%; 25/78). Similar poor returns were reported for outpatients with DM (23.4%; 15/64) and PLHIV (40.9%; 18/44). The reason for not returning the samples was not investigated but the clinical nature of this study reflects the challenges in returning stool samples which can be encountered in routine practice and its burden is often overlooked. Similar reports of low stool sample return have been reported elsewhere.41–43 A study of women invited for a bowel cancer screening programme in England over a period of 6 years showed that one-quarter of patients did not return stool samples despite repeated invitation and resending of faecal occult test (FOBt) kits.41 A systematic review showed that return of samples for FOBt for bowel cancer screening programmes across different countries rarely reached 60%.42 Another paper that studied the barriers to stool sample return for microbiology tests requested by GPs found that embarrassment, hygiene concerns, privacy and lack of information were some of the barriers behind the non-return of stool samples.43 Therefore, it is vital to take these factors in consideration when requesting FEL-1 tests.
One of the limitations to this study was the lack of follow-up to assess the response to PERT, which could strengthen the accuracy of the PEI diagnosis with FEL-1, given that false positives results can occur in low-probability conditions in up to 11% of patients.5 To reduce false positive results with FEL-1, checks were made with patients to make sure there was no contamination with water or urine. We also acknowledge the potential variability of FEL-1 levels in patients. Recent UK guidelines recommend repeat testing with FEL-1 when a cause is not found and, although our cohorts were at high risk of PEI, a repeat test would be recommended with no response to PERT or no evidence of morphological abnormality of pancreas.2 A control group was not incorporated into the current study design; instead, a historical control group was recruited in the same centre and used the same test for PEI; therefore, we felt that it was sufficient to use this group for comparison in our study.
In conclusion, PEI is associated with significant morbidity and can be reversible with enzyme replacement. Our findings suggest testing for PEI in patients with DM or HAI and PLHIV has a significant effect; nevertheless despite evidence to the contrary, our results showed that there is lack of testing in these ‘at-risk’ groups in a clinical setting. Therefore, case-finding practices should be encouraged to detect PEI by testing with FEL-1 in ‘at-risk’ groups. Moreover, future studies with more patients should focus on assessing important end-points in PEI, including assessment of the efficacy of PERT in improving quality of life, symptoms and nutritional markers in ‘at-risk’ groups so that clear guidelines on investigation and therapy can be agreed.
Summary box
What is known?
PEI is under-recognised and can occur in conditions other than chronic pancreatitis, cystic fibrosis and pancreatic cancer. Those conditions are defined as ‘at-risk’ conditions and include DM, PLHIV and patients admitted to hospital with HAI-related symptoms.
What is the question?
What is the current practice and diagnostic yield of requesting faecal elastase tests in patients with ‘at-risk’ conditions?
What was found?
Testing for PEI is low. The diagnostic yield of requesting faecal elastase-1 in patients at risk for PEI was high.
What is the implication for practice now?
Our data suggest that a case-finding approach could improve early detection of PEI in patients with DM, PLHIV and patients admitted to hospital with HAI-related symptoms; however, further studies should be performed to validate this.
- © Royal College of Physicians 2023. All rights reserved.
References
- ↵
- ↵
- Phillips ME
- ↵
- ↵
- ↵
- Terzin V
- ↵
- Vanga RR
- ↵
- ↵
- ↵
- Maconi G
- ↵
- Loser C
- ↵
- Lew D
- ↵
- ↵
- ↵
- ↵
- Burton P
- ↵
- ↵
- Aoufi Rabih S
- ↵
- ↵
- ↵
- ↵
- Martin TC
- ↵
- Yilmaz A
- ↵
- Bhattacharyya R
- ↵
- ↵
- ↵
- Z˙orniak M
- ↵
- Samokhvalov AV
- ↵
- NHS
- ↵
- ↵
- ↵
- ↵
- Dandona P
- ↵
- ↵
- ↵
- Dozio N
- ↵
- ↵
- ↵
- ↵
- ↵
- Dominguez-Munoz JE
- ↵
- ↵
- ↵
- Lecky DM
Article Tools
Citation Manager Formats
Jump to section
Related Articles
- No related articles found.