Subjects
The subjects were patients aged ≥ 20 years old in the period from April 2019 to June 2020 who agreed to participate in the study. All subjects were inpatients and were classified into those with PC, CP, other pancreatic diseases, and a normal pancreas. PC cases were histopathologically diagnosed with pancreatic duct cancer by surgery or EUS-guided fine needle aspiration. CP was diagnosed using the M-ANNHEIM criteria [12]. Patients admitted for a disease other than pancreatic disease and in whom pancreatic disease was excluded based on imaging and blood tests were included as subjects with a normal pancreas. The exclusion criteria were pregnancy; patients being fasted long term; use of antibiotics, probiotics, or pancreatic enzyme replacement drugs within one month before the test: history of surgery on the digestive tract or lung; presence of concomitant disease of cancer of other organs, stage 2 or advanced chronic renal failure, decompensated cirrhosis, active pulmonary disease, gastrointestinal obstruction, apparent gastrointestinal hemorrhage, caries being treated, or periodontal disease; no written consent, and judgement as inappropriate by a physician in charge [8, 9].
Pancreatic function test
The BT-PABA test and the 24-hour urinary C peptide excretion (CPR) test were performed as exocrine and endocrine pancreatic secretion tests, respectively. These tests were performed under non-fasting conditions within one week before and after measuring breath hydrogen. In both tests, the measurement was repeated 3 times on different days and the mean was used for analysis [13]. A PABA excretion rate of < 73.4% was regarded as reduced exocrine pancreatic secretion [14], and a CPR rate of < 29.2 µg/day was regarded as reduced endocrine pancreatic secretion.
Breath sampling
All patients ate a hospital meal on the day before the test and were fasted after 21:00 with drinking of water only. On the day of each breath test, the patients brushed their teeth at 7:00 a.m., breathed deeply twice, and held their breath for 15 s while end-expired breath was collected. Cigarette smoking, alcohol intake, excess exercise, and eating between meals were prohibited after admission. The collection method followed that of the Rome Consensus Conference and North American Consensus in 2017 [8, 9].
Expired gas analysis
A sensor gas chromatograph (SGHA, Nissha FIS Inc.) was used for measurements. Analysis of the results was performed using specialized SGC Analysis Software. Unlike general GC analysis, quantitation was performed using the peak height (= signal intensity). Hydrogen, carbon monoxide, and methane were measured and the target measurement level was 1.0-100 ppm.
Sample collection of microbiota and 16S rRNA gene sequencing
Feces collected during hospitalization were rapidly frozen. DNA was isolated from feces using a DNeasy PowerSoil Kit (Qiagen, Hilden, Germany) and amplified by targeting the V3–4 region of bacterial 16S rRNA using universal primers (forward: 5′-TCG TCG GCA GCG TCA GAT GTG TAT AAG AGA CAG CCT ACG GGN GGC WGC AG-3′ and reverse: 5′-GTC TCG TGG GCT CGG AGA TGT GTA TAA GAG ACA GGA CTA CHV GGG TAT CTA ATC C-3′). The PCR products were pooled, and sequencing libraries were constructed and sequenced using an Illumina MiSeq sequencer. Pair-End Reads were prepared using MiSeq Reagent Kit v3 with 2×300 reads and 600 cycles (Illumina, San Diego, CA, USA). Analysis of 16S rRNA gene sequence data was performed using USEARCH 6.1, Microbial Ecology (QIIME 1.9.1) and Greengenes v.13_8.
Analytical methods
For between-group comparison based on the BT-PABA test, the subjects were classified into PEI and non-PEI groups based on the criterion of a PABA excretion rate of 73.4%. Associations of the PABA excretion rate were examined with age, height, body mass index (BMI), pancreatic disease (PC, CP, and other pancreatic diseases), rates of concomitant diseases (hypertension, dyslipidemia, diabetes), history of alcohol intake, blood test findings (hemoglobin, HbA1c, creatinine, urea nitrogen, amylase, lipase, total protein, albumin, CEA, CA19-9), and presence of characteristic imaging findings in pancreatic disease (pancreatic hypertrophy, calcification, pancreatic cyst, main pancreatic duct (MPD) stenosis, and MPD dilatation). Concomitant diseases were defined as follows: hypertension, ≥ 140/90 mmHg blood pressure or treatment with an oral hypotensive drug; dyslipidemia, ≥ 160 mg/dL LDL cholesterol or treatment with an LDL-lowering drug; and diabetes, ≥ 6.5% or higher hemoglobin A1c or under treatment [15, 16]. Regarding alcohol ingestion, a subject with a history of continuous ingestion of ≥ 80 g pure ethanol a day [17] was defined as a heavy drinker. Pancreatic hypertrophy was defined using the criteria of Haage et al., in which the thicknesses of the pancreatic head and tail correspond to one or more vertebral bodies and 2/3 or more of the vertebral body, respectively [18]. A MPD with a diameter > 3 mm was regarded as MPD dilatation. MPD stenosis was diagnosed using endoscopic retrograde cholangiopancreatography (ERCP), magnetic resonance cholangio pancreatgraphy (MRCP), and endoscopic ultrasound (EUS), and calcification and a pancreatic cyst were diagnosed using computed tomography (CT) and EUS.
The relationship between endocrine pancreatic secretion and fasting expired gas levels was investigated based on the associations of FBHC, fasting breath carbon monoxide concentration (FBCC), and fasting breath methane concentration (FBMC) with pancreatic function. For exocrine pancreatic secretion, intestinal bacterial flora were compared between the PEI and non-PEI groups.
Statistical analysis
Statistical analysis was performed using SPSS v.27.0 (IBM Corp.). All tests were 2-sided and P < 0.05 was regarded as significant. Continuous variables were analyzed as the median and range. Comparison of data that were not normally distributed was with a non-parametric Mann-Whitney-U test. Differences in rates between two groups were examined by Fisher exact test. Correlations between expired gas levels and pancreatic function tests were analyzed using a Spearman correlation coefficient (r). The cut-off value for FBHC for diagnosis of PEI was determined from a receiver operating characteristic (ROC) curve and the area under this curve (AUCROC), so as to maximize the Youden index (sensitivity + specificity-1). For between-group comparison of intestinal bacteria, LEfSe (http: //huttenhower.sph.harvard.edu/galaxy/) was used with default settings.